Is there an association between video games and aggression?

It was a headline in the Daily Mail that started it. “Computer games leave children with ‘dementia’, warns top neurologist”. It was annoying, because (a) there’s no evidence that games cause dementia in kids, and (b) the top neurologist wasn’t a neurologist. Scaremongering stories about the clear-cut negative effects of video games crop up in the news far too often, but when you start to dig into the evidence behind the claims, the story becomes murky. So rather than simply moan about the problem, Suzi Gage and I, along with some colleagues from the University of Bristol and UCL, decided to do some research for ourselves.

A few years later, and the fruits of our labour have just been published in PLOS ONE. Using data from the Children of the 90s study, we set out to answer a (seemingly) simple question: is there an association between playing violent video games at young age, and aggressive behaviour during teenage years?

Actually, it’s not a simple question at all. Before you can attack it in any sort of meaningful way, you first need to figure out two things: what do you mean by ‘violent’ video game, and what do you mean by ‘aggressive’ behaviour. This is a real sticking point in the research literature – as we’ve talked about before, aggression tends to be poorly defined, and in a way that doesn’t map on to real-world violence. In some experimental studies for instance, the way aggression is measured is so poorly implemented that you can essentially find whatever association with video game play you want, simply based on how you analyse the data. But the question of what we mean by ‘violent games’ is also an important one to think about – Call of Duty and World of Warcraft have both been considered to be ‘violent’, but the actual gameplay mechanics of each are very different.

So in our study, we tried to take a more considered approach to these issues. First off, we used a clinically validated assessment tool to look at two aspects of negative mental health: depression, and conduct disorder at age 15. Conduct disorder is a behavioural disorder that encompasses a range of ‘bad’ behaviours, from bullying and fighting, to being cruel to animals, vandalising property and stealing. In other words, rather than relying on a proxy measure of aggression, we looked at whether children were actually engaging in disruptive and violent behaviour. From the Children of the 90s data then, we used a banded measure of conduct disorder – a 6-level variable based on the probability of having the disorder, ranging from (1) unlikely to (6) very likely.

The ‘violent’ video game issue is a bit more complicated. While we were lucky that the Children of the 90s dataset had some measures of game use, they weren’t as detailed as we would ideally like. In the study, children at age 8/9 years were asked what type of computer games they had at home, and the answers were based on broad game genres. Critically though, there was no ‘first person shooter’ category - the only one that came close was ‘shoot-em-up’. So we had to make an assumption that if kids were playing games like GoldenEye 64, they would have ticked the shoot-em-up box. It’s by no means perfect, but it’s the best we could do with the data.

We then identified a number of potential confounding factors - things which might have an impact on any potential association between game use and aggressive behaviour that we would need to take into account. These included factors like family history of mental health problems, maternal education and socioeconomic status, and whether the child at age 8 was a victim of bullying or had emotional problems.

So, the results. Taking into account all of the above factors (and others), we found a weak association between the genre of game played at age 8/9, and conduct disorder band at age 15. This is a bit of a nuanced finding, so let’s break it down a bit. We compared three different ‘genres’ of game – shoot-em-ups, puzzles, and ‘does not play games’, and found that, per categorical change in ‘genre’, there was a 19% increase in risk of being in a higher conduct disorder band. Note, that’s not the same as there being a 19% increase in the chance that you will be diagnosed with conduct disorder. Plus, the absolute risk of being diagnosed with conduct disorder was small – 26 out of a total sample of 1815 children met the case status for the disorder. So in other words, there is an association between playing shoot-em-up games at a young age, and later aggressive behaviour. It’s just a weak association.

One issue that crops up in the research literature quite a lot is that it can be misleading to compare ‘violent’ games like first person shooters with ‘non-violent games’ like puzzle games, because the two categories differ from each other in ways other than the presence or absence of graphic content. Call of Duty is a pretty competitive, fast-paced game, whereas something like Candy Crush Saga is more easy-going and solitary. This was equally an issue for our study, so we tried to take this into account by running a sensitivity analysis.

We looked at data from children who reported that they selectively played shoot-em-up games, and compared them to children who selectively played competitive racing or sports games. Again, there was only weak evidence here that playing shoot-em-up games was associated with an increased risk of conduct disorder, compared to competitive games. So we weren’t able to rule out the possibility that it’s actually competitiveness in games that drives any link to aggressive behaviours – an idea that has been considered in the experimental literature before.

So, putting all of this together, what’s hopefully obvious here is that our study doesn’t paint a clear-cut picture of the effects of playing ‘violent’ video games – the association in fact seems to be quite complex. Of course, one study can never paint a definitive, conclusive picture, and this one is no exception. Instead, we wanted to see it as more of a starting point; a way to kick-start some discussions of how to make the research area more robust and, well, useful.

Our study isn’t perfect; using genres to categorise game violence really doesn’t work well at all, and if we want to really figure out what’s going on here, we need to start thinking hard about what specific aspects of game content might be important to look at in future research. We also didn’t really look at the context in which games are played, and this is something that I’ve droned on about before. Some ‘violent’ games are competitive. Some are cooperative. Sometimes you might play them on your own; other times you might have some friends around for a games night. These sorts of distinctions are important, because without considering them, we’re probably asking the wrong sorts of research questions.

And if we’re asking the wrong questions, ultimately, we’re not really finding out anything useful about the world.