Skeptoid PodcastSkeptoid on Facebook   Skeptoid on Twitter   Skeptoid on Spotify   iTunes   Google Play

Members Portal

Support Us Store


Get a Free Book



Use Caution When Reading Anything About Guns

by Eric Hall

June 14, 2014

Share Tweet Reddit

Donate I had reservations about writing this post. I have considered writing it a few times, only to put it away because it is a difficult subject to tackle in a single post. The subject matter is also highly politicized and very personal. But after some recent news stories and the usual politically slanted postings coming through my news feeds, I decided to tackle a small portion of the gun debate. I am going to make my best attempt at putting this through a skeptical lens, which is my aim in writing this.

Before I chat a little bit about the data, I want to reveal my own bias. I do own guns. I am a hunter and this is my primary use for them. Though I do enjoy target and trap shooting, I rarely have time for those things. I also will reveal that I don't think further action politically will have a great impact on any deaths associated with guns. This is my current assessment of the available data"which I will admit is weak and disjointed. I am also wary of any government solutions when it is not clear, via data and research, whether such a solution is the best or only one. I will do my best to not bring my bias into this post, but because of the nature of the subject that may not happen entirely.

My overall assessment is we have a lot of data. The Centers for Disease Control, the Bureau of Justice Statistics, and to a lesser extent private groups, all have research into deaths and injuries specifically attributed to guns. Interestingly, there is a gap in some of the CDC data due to the CDC not doing gun research for a 17 year period. Even in this the political agendas are evident. There are news stories saying President Obama lifted this "ban," though technically there was no ban. Congress had cut funding in 1996 specifically for the amount the CDC spent on gun research, but there was no actual ban. What does remain, however, is a gap in their data.

One problem I noticed with the data is that even with the amount available, studies typicallyonly focused on gathering and not on connections. Looking through the FBI and BJS data, there are plenty of charts thattalk about violence, homicides, firearms, race, age, and all sorts of categories. But none of these statistics include preceding data, only post event data. Every violent crime event gets a full set of data as to who committed it,where it was committed, and who the victims were. Data on what led up to the crime, which admittedly would be harder to collate, arenot readily available. This I think would be important if looking for a way to reduce violence.

Another problem is the data are often complied and reported by those either carrying a bias, trying to get ratings, or both. An example comes from some recent school violence which made national headlines. A recent school shooting (I won't link a story here or mention the name because I am not a fan of the excess media attention given to these events) brought out a piece of data provided by a gun control lobby group which stated there have been 74 school shootings since December 2012 when another, larger school shooting happened. But as it was quickly discovered, this was a map basically of any homicide by firearm where the word school was related, such as:
Includes an Aug. 15, 2013, incident in which police were called to a high school parking lot at 2 a.m. in Clarksville, Tennessee, where they found the body of a 38-year-old homicide victim with no links to the school.

In another 2013 incident on the list, a 23-year-old Morehouse College student in Atlanta was shot and killed. His body was found near the college, according to news reports, but not on campus.
CNN worked to correct this number and they came up with a much lower number of 15. The Oregonian did a good job of looking into how both numbers were derived and found neither CNN nor the lobbying grouphad good definitions for what qualified. They first defined their criteria in a specific manner, using these criteria:
In how many of the cases did students or faculty face imminent danger from an armed assailant on or near school grounds?
In that definition, they estimate the number to be about 35. They admit that some events in that setinclude evidencethat the violence wasgang-related, according to police reports. This is a much better way to look at data"by being clear in definitions and also explaining the limitations to those definitions.

Related to the school violence numbers involving a firearm is the amount of confirmation bias that appears when discussing school firearm violence. Schools should be safe for our kids. They should be places where childrencan explore, learn, and grow without fear of being harmed. And when that image is damaged by an incident, it brings about an almost visceral reaction that something must be done. Is it getting worse? Let's look at the list of total school homicides, ages 5-18, from the 1992-1993 school year through the 2009-2010 school year: 34, 29, 28, 32, 28, 34, 33, 14, 14, 16, 18, 23, 22, 21, 32, 21, 17, 17. These are all homicides at school, not justthoseinvolving firearms. For all of these years, it varies between 0.9-1.8% of all homicides in that same age group. In other words, only a very small portion of homicides actually happen at schools when compared to the overall number of homicides. Onemight point to other mass shootings or the fact that some victims are not killed in these incidents"which is a valid point. The data here isn't quite as easy to obtain. A couple of other pieces of data give a partial figure. Over half of all violence, both with and without firearms, happen in or near the victim's home or a relative's home. Second, many acts of violence that involve a gun don't result in physical injury;there might not be a single shot fired. So the best piece of data I can find here is this statistic: during the 2007-2011 period(the most recent completed data), there were an estimated 520,094 non-fatal violent incidents in the U.S, with 150,761 of those happening at school. But, looking atthose with firearms involved, there were 107,331 total events with 6,541 happening at school.So there is quite a bit of violence at school, but very little of that violence happens with a firearm. Much more of the firearm-involved violence taking place happens elsewhere. This doesn't tell us the ages of those involved.

Overall,homicides of those 5-18 during the 2008-2009 school year was 1571. This is roughly half of what it was during the 1992-1993 school year. But it is still 1,571 dead kids by homicide. As I pointed out to Sean Malone, in 2010 the number of 12-to-17-year-old people killed in a firearm homicide was 708. He rightly points out that that number has decreased every year (which is good), but then used the comparison of just that age group to the entire U.S. population. He pointed out thatbecause those deaths are only 0.000226% of the total U.S. population, it is an outlier and should not worry us. Sure, the data say the chances of any one of our children being killed by homicide using a firearm is small, but should that end the conversation? I pointed out that cystic fibrosis has a death rate of 0.00022% of caucasians. (I couldn't find race-combined data, mostly because CF affects caucasians at a much higher rate than other races.) But we still concern ourselves with CF. Using percentages isn't always the best way to describe data. I can say product X increases your chances of getting cancer by 100%, but if the chances go from onein a million to twoin a million, that 100% doesn't hold much meaning. I don't think we should ignore well over 1,000 yearly deaths by firearm homicide just because it is a small percentage. That does not imply the firearm is to blame and that gun control is the answer. I simply don't like using data in this way.

A hard part about looking for links between the data and finding solutions is this is much more a social and psychological problem than many things skeptics normally tackle. In the case of vaccines, for example, the last couple of decades have shown only a couple hundred (or less) pediatric influenza deaths each year. The data arevery clear that the more people that are vaccinated, the fewer deaths and injuries there are due to those diseases. But should we be so harsh on people when it is only a couple hundred deaths? Of course we should, because we know the cause and we know the solution to preventing most of those deaths. However, just because the cause and solution for these firearms deaths isn't clear doesn't mean we should negate those deaths or not look for a cause and solution (hopefully in that order).

(Please excuse the large generalities here for a moment. I know we can all try to pull tiny trends out of the next data I am going to mention, but let's not fall in that trap. I am going to make a few general observations and state again the data don't tell us much about cause and effect. The only thing we have is that the data aretemporally related.)

We don't have a great guide so far on gun control being a solution. In England, where virtually all firearms are banned, violent crime hasn't really decreased at all. In Australia, a fairly strict version of gun control passed and is correlated with a drop in crime involving firearms. In the U.S., violent crime has gone down in the last two decades, including through a period of less gun control as some measures expired in that time frame. Often cited by gun rights advocates is Chicago, which despite its near total ban on guns has not seen a decrease in crime. I again think this is a misuse of data, as a ban in such a small area surrounded by a much larger area without the ban is not a good comparison. I compare this to Minnesota having no Sunday liquor sales. I am sure people living in central Minnesota don't buy much in the way of alcohol on Sunday. But if one lives on the border with Wisconsin, it isn't difficult to get access to liquor on Sunday. Chicago, to me, is not a good example to use for the effect gun control has on violence. Because control laws are inconsistent, and becausethe data act inconsistently with control laws in various countries, I don't think anyone could reach a good conclusion to the effect gun control has on violence. And that's for both sides: data of violence going up or down after laws are put into effect is not clear. Let's not use weak data to make very large decisions.

Anotherclaim thatmany anti-science people and even the media pushes on the public is the correlation between antidepressants and firearm violence. After certain school-violence events, I've seenreports of the assailants having beentreated withsome sort of antidepressant or ADHD medication. Because of the black box warning these drugscarry, the assumption is that the medicationsdrove them over the edge. Because we still have much to learn about brain chemistry, I don't think we can definitively answer the question if these medications are contributing to these incidents. But we have a couple pieces of data to compare. Over the last couple of decades, the number of available antidepressants and the number of prescriptions for those drugs has increased markedly. In that same period, violent crime in general, including with firearms, has gone down significantly. Other studies even show a decrease in suicide rates as prescriptions have gone up. So even if these drugs caused a chemical imbalance which drives people to violence, the net effect could be to prevent violence more than it causes it. Air bags actually cause a few dozen deaths every year, but they also save thousands. We should always take the net benefit when looking at the data.

Some activists will try to blame video game violence. I don't have a good way to measure the level of violence in a game thatwould qualify it as a game capable ofdriving someone to commit an act of violence in real life. But, we do have overall sales. From 2000 to 2008, video game sales went from $20 billion to $50 billion in the U.S. In that same time, all violence, including with firearms, and in all age groups fell. This, like the antidepressant case, is a correlation. While correlations are helpful, they are not generally definitive, especially when confounding factors are difficult to control. Again, in highly publicized shooting cases, the shooters are very often into video games with heavy violence and shooting. But withthe overalldata one could concludethat these outlets for aggression prevent violence"at least with the same weight that activists use to say they cause violence because of the correlation that shooters often play these games.

Other causes for the reduction in crime have been proposed, but are not easily verified. For example, the bookFreakonomics (2005) takes on a study,published by one of the authors, in whichthe legalization of abortion is correlated witha reduction in crime. It has recieved some criticism in its methods and data analysis. Another set of studies have correlated the removal of lead from gasoline and paint as a cause for the reduction in crime. The interesting thing is in both casesis atime delay thatmatches pretty well. Is it both? Is it neither? This is not an easy thing to determine.

In the end, the hard data available attest thatviolent crime, including with firearms has gone down on a pretty consistent basis. Overall violence has also gone down. The amount of school violence has gone down in similar proportions. The amount of non-firearm violence far exceeds violence involving a firearm. Most firearm violence doesn't involve shooting, but instead the presence of a firearm. This reduction trend is one we should continue and accelerate if we can.

With these hard data and some hypotheses as to why this reduction is happening, we should continue to study violence, the data gathered, and the reason for the reduction. We should see if we can further reduce violence and accelerate it if possible.

I wrote this post because I don't like misuse of data, especially as a way to shut down the conversation about violence, including firearm violence. Firearms hold auniqueplace in the minds of many, both those for and against their use for any reason. The social construct is one part of the conversation. It is not a reasonable position tosay"the odds of someone getting shot is so small"less than dying in a car accident"so no need to do anything." The odds are not zero and there are thousands killed by firearms each year. It is not a reasonable position tosay"one school shooting is too many so ban all guns." Guns have legitimate uses for sport and have been used for protection. If we banned every object with the potential to cause harm we would be without objects. There is a risk reward for everything. Using labels like "assault weapons" or "gun violence" only confuses the issue and doesn't match the data or reality. Whenever something is framed in the aspect of fear, we shouldn't react with fear. Instead, we ought to ask "Why are they trying to frighten me?" (Thanks to fellow blogger Craig Good for this question)

I want the conversation to continue because something must be working"violence continues to decline. My concern is that decisions based on fear and not on good data and good science could actually reverse the trend, even if the intentions are good. The numbers show we still have a very large violence problem. As a recent incident insouthern Minnesota highlighted, it doesn't require a firearm, or even an occurrence of violence to cause both fear and harm. (Astudent left a small explosive device on school grounds that didn't go off,but he had a much larger cache of explosive material in a storage unit, which was volatile and could have damaged a significant area around the storage unit if it had detonated.) The data I see doesn't lead me towards any single cause or solution. We need to be aware of our own biases, examine the issue, decide on what research should be done next (hopefully with reduced bias as well), and make measured decisions based on science and not on emotion. That is what scientists and skeptics push for in all areas of study, including social ones such as this.

I wish I had something more profound than that. But that's it. This issue charges people's emotions, but what the data tell us is that something is working. Let's try to find the thing or things that are working. Science can help us with that. We have good medications, but we don't stop looking for better ones using science. We have safer cars than in the past, but we keep looking for new ways to make them even safer using science. We are reducing violence, let's look for ways to improve that using science.

I know guns and gun control can be a very controversial and personal topic. Please keep the comments on the subject and not attack each other personally. Each blogger on Skeptoid policestheir own comments, and in general we allow wide latitude for comments of various viewpoints. I do reserve the right to delete comments that are overly offensive or do not contribute to the conversation and constitute a personal attack. Thanks!

by Eric Hall

Share Tweet Reddit

@Skeptoid Media, a 501(c)(3) nonprofit








Want more great stuff like this?

Let us email you a link to each week's new episode. Cancel at any time: