Monday, February 15, 2016

Shooting rampages increase the risk for research into violent video games

“This just in:  Apparently, there is more than one connection between all of the previous rampage killers in the US. Not only did they all play violent video games at some point, they all watched TV news AND had access to guns.”

While the above discussion is obviously fake, according to footage in the Colbert Report of September 18, 2013 one TV journalist has suggested that we could perhaps prevent future firearm violence by keeping a register of video game purchases.

Correlation problems
xkcd on reverse causation


 How did the discussion of what is behind these horrible shooting rampages end up being about video games?  Simple:  As a society, we want to understand what caused this so we can prevent it in the future. 


“We don't have a lot of control over many of the factors that can contribute to violent behavior. But we have some control over violent video games. We can make it more difficult to get access to them. We can strengthen our laws against teens acquiring these games.

Hold the phone; is he talking about video games or guns?  Let’s rewind and make a substitution:

“We don't have a lot of control over many of the factors that can contribute to violent behavior. But we have some control over guns. We can make it more difficult to get access to them. We can strengthen our laws against teens acquiring these guns.”

Everyone knows that you can’t have firearm violence without firearms. The cool thing about public health is that
  • there is (ideally) a scientific approach to understanding the processes that lead to outcomes, and
  • there is also (ideally) an approach to deciding what is important to study and regulate. One of the key questions to ask about a potential research subject is:  how much of an effect would being able to change this potential cause have on my outcome of interest?

So,  keeping in mind the important limitations of public health research I discussed in my earlier posts, let’s talk about just a couple of the reasons why keeping video games on the research and policy agenda is a waste of time and resources:

1.       Guns are clearly responsible for firearm violence and can be regulated, even in other countries that have traditionally refused limitations. (Take a look at the Emmy Award winning segment of The Daily Show, where comedian John Oliver discusses the complete absence of shooting rampages in Australia after gun control with former Australian prime minister John Howard.)

2.       Scientists have been studying the effects of video games for decades, and there is still no consistent relationship between violent video games and real-world violent acts. In fact, one recent well-controlled study showed no long-term ties between violent video games (when measured by themselves) and real-world seriously violent acts when other factors like substance use or living in a violent community are taken into account (p.932; remember the previous discussion about statistical levels and coincidence). It did show, however, that visiting websites that “feature real people fighting, shooting or killing” is associated with real-world violent acts. I’ll get back to more of the methodological problems in future blog posts, but feel free to read anything by Chris Ferguson or Craig Anderson for more information and contrasting views on the research to date.

3.       Video games and other media exposures are not proximal causes, only distal links on a chain of other causes (see an excellent article by Krieger for a thorough discussion).   And contrary to the implication above, we actually do have some control over the more proximal causes (e.g, guns).

4.       Multifinality and equifinality. These limitations lead to different ways of viewing how likely or probable something is (risk). And this little bit is going to be our public health lesson for the day.

Multifinality:  People can have the same risk factors (potential causes of a bad event) but different outcomes.

Aaron Alexis and Adam Lanza both played video games (like the majority of people in the US). They also (probably) watched TV, were male, and walked upright.

The above helps us understand questions about risk that are vital to approaching things from a public health perspective. For the terms below, I paraphrase a discussion by Swanson, substituting “violent video games” for “mentally ill”.

Absolute risk:  The vast majority of people who play violent video games are not violent.
Relative risk:  People who play violent video games are no more likely to commit violent crimes than those who do not play violent video games.  (However, people who have access to guns are more likely to shoot people.)

And then there’s equifinality:  A variety of risk factors can lead to the same outcome.

Aaron Alexis was a former Naval reservist with a criminal history of violence and current money troubles. He was also an African-American practicing Buddhist who had sought help from the Veterans Administration for paranoia and hearing voices.  He had a legally-purchased shotgun.

Adam Lanza was a well-off young white man who lived an isolated life after years of being bullied in high school. He also had access to a stockpile of guns with over 1600 rounds of ammunition.

Attributable risk:  Violence is a societal problem caused largely by things other than violent video games (ready availability of guns, for example).

The stories we hear about rampages are just that, stories. They may serve the purpose of trying to make sense of something horrible, but it’s time to move on from investing more time and money into studying video games and consider what could provide “more bang for the buck”.  Some of the key assumptions of the Code of Ethics for Public Health say that a base for action should improve health through seeking and acting on knowledge, and that there is a moral obligation to share what is known.


Yes, there are barriers to getting the knowledge out there (e.g., publication bias, selective outcomes reporting). Yes, it is hard to take action on things that we know (e.g., gun control politics). But given that toddlers with guns seem to have killed more Americans this year here in the US than terrorists did, it’s time to take advantage of this window of opportunity and enact real change in the clearest cause of firearm violence—access to guns.

reprinted from mcoldercarras.com, 9/2013

1 comment:

  1. This also illuminates the difference between correlation and causation. It's also so sad to see America's love affair with guns continuing with such great adverse consequences...

    ReplyDelete