Does Gun Control Work?
Natural experiments suggest YES
Posted Mar 19, 2018
Like so many of us, I’m outraged at the prevalence of senseless killings and injuries by gun violence in the United States. And I have not been quiet about it. As I’ve written about in prior posts, I think it is important for people to understand how overwhelming the scientific research is showing a connection between the prevalence of guns owned and the number of gun-related deaths and injuries.
Warning: The Rest of this Post is Kind of Technical
In the behavioral sciences, there is an important distinction between research that we call “correlational” versus research that we call “experimental” in nature. In short, correlational research is characterized by research in which you measure variables as they exist naturally - without you, as the researcher, manipulating them. A classic example would be if you were to study the relationship between socioeconomic status and educational attainment. A positive correlation has been regularly documented between these variables. But researchers on this topic have not manipulated either of these variables. In other words, researchers haven’t randomly assigned some participants to be wealthy and others to be poor - and then examined differences in the relative educational levels of those two groups. That would be, of course, completely unethical to do!
With “experimental research,” you carefully control one of your variables (the “independent variable”) by manipulating it, to see if the different levels of that independent variable cause changes in some outcome variable. For instance, if you randomly assign half of your participants to drink 5 cups of coffee and the others to drink no coffee and then see how these two groups differ on a measure of anxiety, you would be able to see if coffee actually affects (causes changes in) anxiety levels.
In the behavioral and social sciences, the rule is essentially this: If your research is experimental in nature, you are able to draw inferences regarding a “causal relationship” between your variables. If your research is correlational in nature, you are only able to talk about a “relationship” between your variables, knowing full well that a simple relationship between two variables does not necessitate that one of the variables causes the other. This is simply the nature of how our research methods work.
Sometimes, You Cannot Do Experimental Research
All things equal, it’s better to say that you have found a “causal relationship” between two variables than just a “relationship” between two variables. However, sometimes practical and/or ethical constraints make it impossible to conduct an actual experiment on some research question.
Research on gun violence is, in fact, a classic example of such an area. To truly see if the prevalence of guns cause reduced levels of gun violence, you would need to manipulate the prevalence of guns within a population by random assignment to conditions. That means that you would have to choose some participants, by some random process, to own guns and you would have to randomly choose others to necessarily not own guns. See the problem? Imagine actually implementing THAT research!
Because of this practical issue, research on the effects of gun ownership and gun laws on gun violence is essentially necessarily limited. And such research is particularly limited in its ability to make causal links.*
Natural Experiments as a Workaround
However, sometimes, due to random world events, something we like to call a “natural experiment” emerges. A natural experiment exists when the creation of randomly assigned groups of people, or the creation of a natural “repeated-measures experiment,” emerges simply due to some events that take place in the world. A “repeated-measures experiment” exists when participants are presented with one situation at Time 1, then some intervention happens, and then these same participants are studied at Time 2 to see if the intervention caused changes in these participants.
Well sometimes, nature works so as to create natural experiments, which come very close to having all of the hallmarks of true experiments. And as such, natural experiments come close to allowing research to draw causal inferences regarding the relationships between variables.
Australia and the UK as Natural Experiments on Gun Violence
1996 was a bad year in both the UK and Australia. Each of these nations had a tragic mass shooting that had qualities similar to what we recently saw in Parkland. In Dunblane, Scotland, a disturbed middle-aged man killed 18 people in a school. The victims were mostly children. He used legally purchased handguns to carry out this horrific act.
That same year, a lone male killed 35 people in Tasmania, Australia using a semi-automatic weapon. Another senseless and unnecessary act of violence that should have been prevented.
The people and the governments of both the UK and Australia were, as you might imagine, outraged - and they demanded change.
Both nations swiftly implemented major changes in their gun policies, with the UK banning all handguns (except for single-shot .22 caliber pistols) and with Australia implementing major gun reform policies including a ban on automatic and semi-automatic weapons. Since these changes have been implemented in these two nations, there has been a single mass shooting in a school in either of them. And it is now 22 years later. On the other hand, thousands of people in the US have died from mass shootings in this same period.
When it comes to the gun issue, the United States is facing a crisis right now. According to the CDC, 30,000 deaths a year in the US are the result of guns (check out a broad array of statistics on this topic at the website of the Gifford’s Law Center to Prevent Gun Violence http://lawcenter.giffords.org/facts/statistics/ ). The causes of gun violence are, of course, many. This said, to suggest that prevalence of guns and lax gun laws are unrelated to this crisis is to turn a blind eye to the scientific data.
Some people argue that because the research on the relationship between gun prevalence and gun violence is correlational in nature, it is then inappropriate to say that the one causes the other. While we are not able to conduct actual experiments on this topic, due to practical constraints, the natural experiments found in history (via the gun reform campaigns of the UK and Australia) actually give us some reason to think that accessibility to guns actually causes gun violence. Does gun reform reduce gun-related violence? Take a look at the history of the UK and Australia on this one - I think you’ll find your answer.
*Importantly, this research is famously limited even more by federal policies such as the Dickey Amendment of 1996 which generally disallows the CDC from the funding of research into gun violence. Note that the Dickey Amendment was adopted by the US federal government the same year that both the UK and Australia moved toward their famously restrictive gun policies.
Acknowledgement: Thanks to all-star New Paltz sociologist, Peter Kaufman, for help developing this post. If you don’t like it, feel free to blame Peter!