Gaming’s Dubious Relationship with Violence

It’s time once again to break down some video game research. Today’s recently published paper explores the relationship between violent video game consumption and violent behaviour. However, it does not do so in a vacuum—it also explores additional factors such as family background and race.

For those who do not wish to read the full breakdown, there will be a summary available at the bottom.

Please enjoy!

Introduction

This paper begins by discussing the less-than-positive societal image of violent video games, including a quote from Hillary Clinton stating that “playing violent video games is to an adolescent’s violent behaviour what smoking tobacco is to lung cancer”. The authors rightfully point out the fallacious nature of this argument with two main points. The first point is that while violent video game consumption is on the rise, adolescent violence is decreasing.

The second point is that video game research is not usually conducted to the standards which permit politicians to make these claims. A number of violent video game research shortcomings are discussed to support this point. An example of these shortcomings is using weak, ludicrous measures of ‘violence’ in these studies (e.g. how much hot sauce a participant gave to someone). Further shortcomings include: how weak the relationship between games and violence tends to be, relying heavily on university samples, and measuring video game consumption as a dichotomy (‘I play video games’ vs ‘I don’t play video games’) rather than looking at rates of video game consumption.

The remainder of the introduction is dedicated to how violent video games could possibly impact young people, such as the idea that they ‘activate’ innate violent tendencies in young people. Two subsequent points are made which relate to the proposed research questions. The first point relates to the potential for frequent violent video game consumption to desensitise young people to the consequences of violence. The second point involves something which the authors refer to as ‘propensity towards violent media’. The argument that is put forth is that young people may have a propensity towards consuming violent media from a very young age due to a lack of parental monitoring and/or care. These points then lead on to the three research questions proposed by the authors:

• Research Question 1: Does time spent playing violent games correlate with violent behaviour?
• Research Question 2: Does the strength of the relationship between violent games and violent behaviour decrease when propensity to violent media is included?
• Research Question 3: Do other social and environmental factors more strongly predict violent behaviour than games?

Method

These research questions were explored using the 2015 wave of the Delaware School Survey. This sample involved 5,133 eighth graders and 3,886 eleventh graders, with 51% of the sample size being female. The ethnicity breakdown of the sample is as follows: 49% white, 21% black, 18% Hispanic and 12% ‘other’ ethnicities.

Violence was measured by asking participants if they had “hit someone with the intention of hurting them” or if they had “take[n] part in a fight where a group of your friends are against another group” in the past year. These questions were answered in a typical ‘Yes’, ‘No’ fashion. This measure of violence is high in what researchers refer to as ‘ecological validity’. Ecological validity quite simply means that it is a valid representation of something which happens in the real world. This is a much more valid measure of violence than getting people to pour out hot sauce as it actually represents violence in the real world. You could argue that young people might be apprehensive about admitting to their past violence, but the methodology specifies that all questionnaires were anonymous and participants could not be identified if they answered honestly.

Before I explain the remainder of the variables that were explored, I would like to describe the type of analysis that was conducted in this study. The analysis used in this study is what is known as a regression. In layman’s terms, the aim of a regression is to try to explain as much of why something happens as possible (known as ‘variance explained’). For example, let’s say I want to look at why people buy ice cream. I would have the variable of ice cream sales, then I would enter variables such as temperature that day, gender, age, yearly income etc. into the regression model. If the model tells me that 34% of the variance is explained, this means that I can say that 34% of the ice creams sold that day were because of the variables I put into the model. It is also important to note that variables can end up not significantly contributing to the model if more variables are added. For example, gender, age and temperature could significantly predict ice cream sales, but if income were added, age may no longer significantly contribute.

Phew. Now that that’s explained, I can describe the variables that will be used to predict why violent behaviour happens. These are:

Violent video game exposure: This was measured using the question “How often do you play video games, such as games that are rated M?”. Responses ranged from “never” to “more than 10 hours per week”. I was a bit apprehensive about this question as ‘violent video games’ is quite subjective – is Mario jumping on a Goomba violent? However, ‘such as games that are rated M’ is a good addition as it helps young people conceptualise violent games such as Grand Theft Auto.

Parental attachment: This was measured using a scale of parental attachment that had a high reliability score.

Youth disclosure: Young people were asked whether their parents typically knew where they were at all times.

Parental enforcement: A measure for parental strictness. This was asked using the question “My parents’/guardians’ rules are strictly enforced’.

Violence at home: Whether young people witnessed violence at home, on a scale of “never” to “almost daily”.

Hit by an adult: Whether the young person was the victim of violence at home, on a scale of “never” to “almost daily”.

Violent media propensity: This is where things start to get really interesting. Remember in the introduction when it was argued that a lack of parental monitoring/care led to people being exposed to violent media at a young age? Well this was something that wasn’t specifically measured in the study. However, this is a variable that was calculated using a number of home and family factors such as socioeconomic status, race, smoking and drinking behaviour at home etc. How many variables to be precise? 171 variables. Some insane calculation went into this variable and it was indeed highly related to violent video game consumption, so hats off to them for all of that work.

Missing data was dealt with by deleting anyone whose gender or video game consumption was not recorded. This led to a final sample of 4,096 eighth graders and 3,117 eleventh graders.

Results

The results of this study were split by three factors: gender, engaging in solo or group violence, and whether the young person was in the eighth or eleventh grade. For the purposes of organisation, the results will be described by gender.

Male Findings

For eighth grade solo violence, time spent playing violent video games only explained 2.8% of the variance in hitting someone. When violent media propensity is included in the violence model, this jumps up to 11% and time spent playing video games stops being significant. In the final model with all variables included, the significant contributors to hitting someone were: violent media propensity, being a victim of violence, and being of a non-white race. High levels of youth disclosure and parental attachment were protective factors against hitting someone. These factors explained 19% of cases where a young person hit someone.

For eighth grade group violence, violent games were not a significant predictor at all and explained less than 0.002% of instances of group violence. The significant variables in the final model were violent media propensity and non-white race, with youth disclosure and playing violent video games protecting against group violence. These variables explained 21% of instances of engaging in group violence.

For eleventh grade solo violence, time spent playing violent video games accounts for less than 1% of the variance in hitting someone and did not significantly contribute to violence. In the final model, violent media propensity, being a victim of violence, and non-white race were the significant predictors of violence, with youth disclosure being a protective factor. These variables explained 21% of instances of hitting someone.

For eleventh grade group violence, neither violent video games nor violent media propensity were related to group violence. Only youth disclosure was significantly related to group violence, and this was as a protective factor. All variables together in the model could only explain around 13% of engagement in group violence.

Female Findings

For eighth grade solo violence, violent video games alone predicted around 5% of violent incidences. It remains significant in the final model, but its influence is very small. The significant factors of the final model also include violent media propensity, arguing at home, being hit by an adult, and non-white race as risk factors, with youth disclosure being a protective factor. This model explains around 22% of instances of hitting someone.

For eighth grade group fights, violent video games start off explaining around 3% of group fights, then stops being significant when violent media propensity is included. The final model explains 19% of group fights and its significant factors are: violent media propensity, violence at home, and non-white race; youth disclosure and parental enforcement are protective factors.

For eleventh grade solo violence, violent video game consumption explains 2% of solo violence, but stops becoming significant when violent media propensity is included. The final model explains 25% of instances of solo violence, with significant factors including: violent media propensity, violence in the home, being hit at home, non-white race, receiving free school meals, youth disclosure (protective factor), and parental enforcement (protective factor).

For eleventh grade group fights, violent video games start off explaining around 4% of group fights and interestingly remains significant not only when violent media propensity is included, but even in the final model. However, violent video game consumption makes the smallest significant contribution to the model. The final model explains 21% of group fighting behaviour; significant factors include violent video game consumption, violence at home, and non-white race, with youth disclosure being a protective factor.

Discussion

As there are quite a number of models to discuss, the majority of the discussion involved explaining and reiterating the findings of the study. The authors use these findings to argue the notion that we need to revise the rhetoric of the damage that violent video games do to young people. When adding factors such as home life, race, socioeconomic status and other factors, video game consumption only contributes to violence models for 2/8 of cases, yet its contribution is always the smallest. It also argues that we need to do further research into households that fully permit access to violent content and media from an early age. Are these dangerous/careless households, or simply apathetic households?

I’ll quickly share some critical thoughts I have on the study:

• Although I appreciate the statistical effort that went into creating the ‘violent media propensity’ variable, I’m a bit apprehensive about how far it can be generalised. The variable is indeed significantly related to violent game consumption, but in order to label it ‘violent media propensity’, we have to assume that it’s also related to watching violent TV and violent movies. I’m not a fan of making assumptions in research, I’d rather use hard evidence.
• I would assume that people will be critical of the generalisability of a sample of eighth and eleventh graders, but it has its benefits for two reasons. The first is that it’s not yet another sample of university undergraduates. The second is that it is actually a sample that people are trying to scaremonger on behalf of. It’s nice that they’re researching the young’uns when that’s where a majority of the concern lies in.
• As usual, it might be difficult to generalise the findings of an American study to countries outside of the US. However, I will say that it is more ethnically diverse than other video game research I’ve seen.

In conclusion, I feel that this is a good study to highlight the relationship between video games and violence where violence is actually well-measured. It demonstrates that we simply cannot look at the relationship in a vacuum. Youth violence is a complex thing, and we owe it to future generations to thoroughly explore it rather than scapegoating for convenience.

Thank you for reading and happy gaming! ♥

 

Summary

• Tired of the general low quality of video game violence research and social rhetoric on the topic, two researchers have explored the link between violent video games and violence where violence is well-measured. This link is also explored in conjunction with other variables such as gender, race and household stability.
• Used data from over 7,000 American youths.
• The impact of violent video game consumption on violence ranged between 0.002% and 5%. When looking at multiple factors, violent video game consumption was related to violent outcomes in 2/8 of scenarios, but its contribution was the smallest in every model. In one scenario, violent video game consumption made young people less likely to engage in group violence.
• The largest contributors to youth violence were being a victim of violence, race and a propensity towards violent media based on household factors and socioeconomic status.