All Gun Statistics are Bullshit

The following is a tentative draft of a chapter of a book. I had intended to wait on its publication until the book itself — and it will be refined, edited, and released in full when that time arrives.

However, given the Surgeon General’s recent declaration that gun violence is a medical concern, it seems relevant and pertinent to add to the argument early.

This post is not my full view on guns, but it conveys the degree to which statistics employed in the gun debate are meaningless (on both sides). Everything Vivek Murthy says is not only dubious and not-credible on its face, but an assertion of philosophical priority, specifically of health above all other values.

This invites a conversation about other values, within the domain of philosophy, and if any of them might, in fact, supersede health in import. This post does not go into that question immediately, but it is meant to show that it is in fact these philosophical questions — and not disagreements about “data” — which guide the gun question.

You’re walking down a street. It’s dusk. A brisk wind blows a newspaper past your feet as you scuffle toward your car.

As you reach out to grasp your door handle, a hand grips your shoulder from behind. You turn around. Leveled at your face is a nickel barreled revolver — staring into your soul like the eye sockets of the grim reaper himself — held in the shaky hand of an addict with nothing to lose, looking for pocket change.

Gun to your head — how do you feel about firearms?

Guns are on our mind.

Some people love guns. They can’t stop thinking about the next firearm or part they want to buy. Others loathe them, from personal bad experience or as a matter of taste, or out of a civic desire to stop gun violence. It’s a touchy subject.

In many ways, guns seem linked to American identity itself, as a symbol of freedom or violence or both — everything good or bad about our country encapsulated in a simple machine. It all comes down to guns.

Legally, things seem pretty clear. The Second Amendment to the Constitution says the right of the people to keep and bear arms shall not be infringed.

But we’ve changed our mind on many laws, including Constitutional Amendments. We’ve added a few. We completely altered the interpretation of the First Amendment (Abrams v. United States, 1919). And we’ve essentially done away with the Fourth, Ninth, and Tenth Amendments, in practice. Would reining in the Second really undo the country?

What if they are right to take our guns away?

Is there a good reason to disarm the population?

The argument over firearms has become superficial. When we aren’t talking about the law, we’re talking about numbers — how guns make us safer, or how they clearly make us less safe. But when the discussion shifts from statistics to logic — from numbers to how you analyze and interpret those numbers — things can get frustrating. It’s as if everything is just a rationalization — a kind of cover-story to justify something that would be less popular if the real idea was stated directly.

It’s important to understand this superficiality if we want to get at what the argument over guns is really about.

Fortunately, this is easy to do. All we have to do is look at basically any statistic pertaining to guns with a touch of objective scrutiny.

In 2022, NPR published a kind of blog-post called “12 stats to help inform the gun control debate”. The statistics it listed were as follows:

100: Number of people killed by guns in the U.S., every day
12: Number of children who die every day from gun violence in the U.S.
950: School shootings since Sandy Hook, including 27 school shootings so far this year.
18-21: Peak ages for violent offending with firearms
8 million: Number of AR-15s and its variations in circulation
9 in 10: Number of people who will die after attempting suicide with a gun
98: Percentage of mass shooters who are men
89: Percentage of gun owners who favor preventing the “mentally ill” from purchasing guns
77: Percentage of gun owners who favor background checks at private sales and gun shows
54: Percentage of gun deaths that are suicides; 43% are murders
79: Percentage of murders that involved a firearm
<1: Percentage of people who defended themselves with their guns in violent crimes

I chose this particular story because when I searched “statistical argument for gun control” on Google, the NPR piece was the top result. Other pages and arguments and statistics like this are everywhere, and the NPR piece is fairly representative of what’s floating around.

These statistics frame the discussion as a matter of safety and preventing loss of life. But upon closer scrutiny, all of them seem to fall apart.

1. “100 people killed every day”

In a country of 330 million people, a hundred a day is a triviality. For context, cigarettes purportedly kill 1,315 per day, alcohol kills about 487 Americans per day, cars take 117 Americans daily, and swimming pools kill 97.

Cancer and heart disease still dwarf all of these… but a funny thing happens when things like “cancer” are brought into the equation. Is “cancer” really different than being killed by a smoking? Sometimes it isn’t, and sometimes it is. If a statistician were to introduce a pie-chart that had separate sections for “smoking” and “cancer,” we might raise an eyebrow. Which one is the “cause?”

How many firearm-related deaths, for example, are more fully caused by alcohol? If someone commits suicide by shooting themselves, would taking away their gun have stopped them from idling their car with a garden hose for 10 minutes? Or jumping off a tall building? Was the gun the cause, or were suicidal feelings the cause, and is putting these deaths as “gun deaths” like calling lung cancer in a smoker “cancer,” as something separate from cigarette related deaths?

Obviously, these two aren’t mutually exclusive. Cancer can be caused by cigarettes. But many non-smokers get cancer — including lung cancer. Many of these statistics emerge from assumptions that are themselves not grounded in data. Perhaps they assume but for the firearm, this person would not have succeeded in killing themselves.

100 Americans a day is high, but it is not that high compared to many other causes, and in fact much, much lower when other causes are given their proper credit.

For example…

10. “54% of gun deaths are suicides”

Shooting is a fairly successful method for committing suicide. However, it isn’t the only successful method, and the assumption that without access to a gun, people would not make some other attempt on their own life is not borne out by observation.

The country with the highest suicide rate in the world is Lesotho (72.4/100,000), where hanging is the most common method. The country with the next highest suicide rate is Guyana, with a rate of 40.3/100,000. In Guyana, the most common method varies by age range: while hanging is still the most prevalent overall, poisoning with pesticide is more common among younger people.

No other country in the world even tops 30/100,000 suicides. But among the more suicidal countries are Eswatini (29.4/100k), South Korea (28.6/100k), Kiribati (28.3/100k), Micronesia (28.2/100k), Lithuania (26.1/100k), Suriname (25.4/100k), Russia (25.1/100k), and South Africa (23.5/100k). The most popular methods of suicide are overwhelmingly hanging and poisoning. Even in the relatively well-armed Lithuania and Russia, hanging is by far the most common means of killing oneself.

At 16.1/100,000 suicides, the United States is tied with Botswana for having the twenty-third highest suicide rate — just below Croatia and above Japan.

A survey of suicide methods around the world was conducted by Swiss psychiatrist Vladeta Ajdacic-Gross and colleagues in 2008. In their findings, only four countries had death by firearm as the most common method of suicide: Columbia (37%), Uruguay (47.8%), Switzerland (33.5%), and the United States (60.6%). Hanging was the most common method in 49 out of 56 nations studied. In Kuwait, Poland, and the aforementioned Lithuania, hanging accounted for more than 90% of successful suicides.

From these broader, less parochial numbers, the idea that reducing gun ownership would reduce suicide is absurd.

2. “12 children killed every day.”

NPR’s linked source for this number is an estimate by Sandy Hook Promise, an activist group that promotes gun regulation. But their estimate likely derives from PEW research data, which argued that gun violence affecting teens had risen from 2.4 fatalities per 100,000 in 2019 to 3.5 per 100,000 in 2021.

But that isn’t all the data said:

In the U.S., some groups of children and teens are far more likely than others to die by gunfire. Boys, for example, accounted for 83% of all gun deaths among children and teens in 2021. Girls accounted for 17%.

Older children and teens are much more likely than younger kids to be killed in gun-related incidents. Those ages 12 to 17 accounted for 86% of all gun deaths among children and teens in 2021, while those 6 to 11 accounted for 7% of the total, as did those 5 and under.

As NPR also helpfully pointed out,

4. 18-21 year old males are the most dangerous

18-21 year old males are by far the most common violent offenders, as well as the most common victims of violent crime.

It would make sense to assume that 12-17 year old males would also be highly represented in this category of violent criminals/victims, since they are demographically close to 18-21 year olds. If we were to break that category down further, we could reasonably expect there would be more 16-17 year olds in this category than 12-15 year olds.

America has a well-established gang-culture, especially in cities. Many of these gangs induce members in their teenage years. In many cases, initiation requires violence. The gangs themselves often get into feuds with each other, with members attempting to kill other gang-members for control of drug and prostitution opportunities, or simply for honor within their group.

Let us assume that of the gun deaths among 12-17 year old males, 70% of those are in 16-17 year olds, and are associated with gangs. This would mean that of the 12 “children” who die every day, just over 60% of these are gang-associated teenagers.

It is impossible to understand the prevalence of gun violence from and to teenage males without this important bit of context. Yet if you search the Pew Research article for the word “gang,” it isn’t mentioned once.

Young gang initiates are not what you ordinarily imagine when you hear the word “children.” You might find yourself thinking of innocent eight-year olds playing in the neighborhood. When these kinds of statistics are paired with stories of Uvalde or Sandy Hook, it feels as if the writers are trying to make you think that that is what they are referring to. But that is not what their own data says.

Speaking of which…

3. “950 school shootings since Sandy Hook” (2012)

This is a baffling statistic. If you go to Wikipedia’s “List of School Shootings in the United States (2000-present)”, you will find that there are in fact only 426 school shootings since Sandy Hook (310, as of the time of NPR’s strange claim).

How they inflated 427 into 950 is unclear. The hyperlink used to cite that assertion makes no mention of that number. It looks as if they simply fabricated the number.

Perhaps 426 (or 310) still sounds pretty high. But even this number is inflated when it is pointed out that the vast majority of these shootings resulted in zero deaths. Indeed, in 2024 alone, Wikipedia already documents 10 school shootings in which there were no deaths and no injuries.

Technically speaking, we could perhaps classify a kid bringing a gun to school and accidentally shooting a light out as a “school shooting.” But it’s not what these highly scientific curators of objective data are implying when they pair reminders of Sandy Hook or Nashville with “shootings” where literally nobody died.

From these 426 listed school shootings, we find only 159 deaths in school shootings since Sandy Hook, over 12 years.

That’s 0.37 deaths per “school shooting,” and about 13 deaths per year.

If one is measuring risk based upon statistics, it might be far more concerning that according to the estimates of the Department of Education itself, “10% of K–12 students will experience sexual misconduct by a school employee by the time they graduate from high school”.

I’ll let you do the math on that one yourself.

11. 79% of murders involve a firearm

You should begin to notice a theme at this point, in terms of how context and information not provided can change the meaning of a statistic.

The country with the highest murder rate in the world happens to be El Salvador, at 52/100,000. This is down from over 100/100,000 in 2015. The government of El Salvador cracked down on gangs, and has seen continual decline in violence as a result.

Other countries with a high murder rate include Jamaica, Lesotho, Honduras, Belize, Venezuela, Saint Vincent and the Grenadines, South Africa, Saint Kitts and Nevis, and Nigeria. The United States comes in with the 76th highest murder rate, at 4.9/100,000. This puts us well below dozens of countries with lower gun ownership and more severe gun restrictions.

Some of this is because people can and do commit murders with bladed weapons (like knives), blunt objects (like hammers), or even their bare hands. In Britain, for example, 41% of homicides were committed with knives or sharp objects. The second highest method was “kicking or hitting” at 19%. Only 5% of homicides were committed by shooting.

It is worth noting that Britain has a relatively low homicide rate generally (1/100,000). But South Africa — which boasts a proud homicide rate of 45/100,000 — also happens to be the current knifing capital of the world with a stabbing mortality rate of 17/100,000. Guns simply aren’t necessary for a high murder rate… nor are guns sufficient to cause a high murder rate, as illustrated by Switzerland, which has an extraordinarily low homicide rate of 0.59/100,000 (about half the rate of Britain) despite having the 14th highest gun ownership rate (28 firearms per 100 citizens).

And let us not forget perhaps the silliest statistic of all:

5. 8 million AR-15s

This one doesn’t need much explanation. According to 2019 crime data from the FBI, there were 10,258 total firearm homicides. Of these, 6,368 were committed with handguns. Rifles – of all varieties, including “traditional rifles” as well as AK and AR variants – were only used in 364 instances.

Armalite-pattern rifles aren’t even close to the most common weapon used in the commission of violent crimes. The “statistic” is a useless bit of trivia. The only reason it appears to be included in this list – which is supposed to “help inform the gun control debate” – is to prime readers to believe that the proliferation of ARs has something to do with violent crime. But this is completely false. They just look scary to people who don’t know anything about guns. Compared to handguns, the criminal use of ARs is negligible.

…at least, that is, if you believe the statistics.

If you’re going to be reasonable, you shouldn’t believe any statistics. Not from NPR, not from me, and especially not polling data which claims that it can tell you what “most people” think – gun-owners or otherwise. The statistics are as misleading in what they leave out as what they say. Where is the data on the relationship between gun ownership and robbery, on the street or at home? Could such a relationship even be measured?

I had a friend who was once followed home from an event by two suspicious men. He noticed them as he was walking down a sidewalk, through the side-mirrors of parked cars along the curb. He happened to be armed. Ordinarily, my friends’ habit was to keep his pistol on his hip. But when he opened his car door, he decided to conspicuously remove his gun from his waist and set it on the floorboard, underneath his seat – all the while, pretending he hadn’t seen the two men following him. Through the mirror, he saw them notice his gun. They promptly turned around and walked back the way they came.

How would one capture and measure things that didn’t happen like that?

The deployment of statistics and data doesn’t actually educate or inform us on issues as complicated as gun control. We might try to use statistics as a sort of weapon, with which to attack the other side. But the argument isn’t really about data. If it was, the writers at NPR would be contemplating strange correlations which don’t fit their story. Take Russia for example. Russia has very strict gun regulation, with only 9 guns in circulation per 100 citizens. Yet Russia has a relatively high murder rate (8.1/100,000) and an extraordinarily high suicide rate (25.1/100,000).

Like Switzerland — but for different reasons — Russia doesn’t fit into the simple idea of how gun regulation will effect violent crime.

Note that this problem with statistics goes in both directions.

The National Rifle Association’s Institute for Legislative Action has produced its own statistical arguments for why gun control doesn’t work. They make the following claims as “quick facts”:

Over the past three decades (1991-2019), violent crime rates have dropped by more than half. The number of privately-owned firearms in the United States doubled in that same period.
The number of people carrying a firearm for protection outside the home has also risen to all-time highs as violent crime dropped.
Mass murderers have repeatedly been deterred or stopped by citizens carrying lawfully concealed firearms.
Concealed carry laws help reduce the number of rapes and robberies overall.

From this, they conclude that “gun ownership is up, and violent crime is down overall.”

To sort of set the tone with statistics, let’s address these in reverse order:

4. Concealed carry reduces rapes and robberies

The citation that the NRA uses here is interesting. The paper — entitled “Confirming More Guns, Less Crime” — appears to be a part of some kind of statistical feud between John Lott, of the American Enterprise Institute, and researchers Ian Ayres and John Donahue. The essential argument of the paper was that other statistical analysts had been misreading data, and that if you group the times differently, there does appear to be some kind of relationship between right-to-carry laws and reductions in violent crime. This would imply that deterrence works.

The feud continued with a subsequent publication by Ayres and Donahue called “Shooting Down the More Guns, Less Crime Hypothesis”, which mirrored the same sorts of allegations back:

While certain facially plausible statistical models appear to generate this conclusion, more refined analyses of more recent state and county data undermine the more guns, less crime hypothesis. The most robust finding on the state data is that certain property crimes rise with passage of shall- issue laws, although the absence of any clear theory as to why this would be the case tends to undercut any strong conclusions.

Now, I’m no statistician. It seems both above and beneath me to intervene and arbitrate in a furious battle to the death between number-crunchers, especially when they seem to enjoy the fight as much as they do.

However, it is worth pointing out that the citation used by the NRA to assert as a fact that “concealed carry laws help reduce the number of rapes and robberies overall” is a highly technical reinterpretation of data which initially suggested that there was no such relationship. When a matter of causal relation is in numerical dispute by nerds over what the proper method of analysis ought to be, it is not a “fact.”

Even if Lott is right, and Ayres and Donahue are completely wrong, there is some marginal relationship over a span of many years, even decades. This data does not — and, in fact, cannot — account for all possible variables which might independently influence measures of gun ownership or legality and violent crime. For example, older populations tend to vote Republican (and with that, generally more concealed-carry laws). Older populations also tend to commit less violent crime. If a population of a given county or state were to collectively age, then we would expect to see both an increased legality for concealed carry and a decrease in violent crime, without the former necessarily having anything to do with the latter.

This is not my assertion, but an example of the kinds of things that can happen while remaining invisible to statisticians who get tunnel-vision.

John Donahue happens to be an interesting counter-example, of a statistician who noticed a relationship between abortion and violent crime.

…Well, maybe. That paper might have neglected a few variables too.

3. Mass murderers have been stopped by armed civilians

This is one of those “probably true but not relevant” kinds of claims, when people are arguing statistics. Statistically speaking, the vast majority of gun-deaths happen in individual altercations, whether it’s a robbery or a crime of passion or a gang conflict.

Mass shootings catch our attention, but simply following your feelings there is an un-statistical way to view the world… though perhaps if we cared about mass shootings specifically, John Lott might be correct (yes, it’s Lott again).

2. Carrying guns has reached all time highs while violent crime has dropped

You won’t believe who wrote the article that the NRA used for their citation…

Interestingly, this correlation-insinuated-as-causation hides a second correlation-as-causation inside of it, like a kind of Russian Nesting Doll of bad logic. In Lott’s paper on concealed carry holders, he says that “[p]ermit holders continue to be extremely law-abiding.”

For those who aren’t aware, you can’t apply for a concealed carry license if you are a felon. As a general rule, criminals tend to be prior offenders — crime predicts crime. What you have in the population of concealed-carry holders is a self-selected distillation of non-felons. Of course they are going to be law-abiding compared to the public when the public includes felons. But that doesn’t tell us anything meaningful.

Now it does appear to be true that carrying guns has reached new heights in the United States. It might also be true that violent crime is on the decline. On the surface, this might feel like it implies that more people carrying reduces crime. At the very least, it proves that more guns doesn’t mean more crime… right?

Well, maybe. In all likelihood, that’s probably true. But it depends on how the data is broken down. I mentioned age before — it turns out that America has been getting older, with an average age of 38.9 in 2023, compared to an average age of 35.3 in 2000 and 32.9 in 1990. We could imagine all kinds of other reasons why violent crime might be going down, independent of concealed-carry. Maybe cyber crime pays better now.

This brings us to,

1. Violent Crime has dropped by half while gun ownership has increased

It’s a subtly different version of the same point. But what about this massive drop in crime, and its relationship with guns?

Interestingly enough, the decline in crime didn’t just coincide with more guns in America. It also coincided with more restrictive gun laws in Europe, where the decline in violent crime not only matched but surpassed the decline in America.

So not only did more guns not cause the decline in violent crime — one could actually make an argument that maybe we could have brought crime down even further if only we had followed Europe’s example and restricted firearms more heavily.

It’d be a bad argument, but you could make it… at least until someone pointed out some other country that restricted guns and still got the same or more violent crime, like Mexico or Brazil.

All of that tedium is to say that none of this is about statistics.

The truth is that guns make some situations better, and others worse. If a woman is being chased by a strange man, things would be worse if the man had a gun. If she had a gun, things would be a bit better. Alcohol, tough relationships, high-crime areas, dangerous wildlife, the responsibility and maturity of the individuals involved, and many other factors play into whether a given firearm is beneficial or costly, and to whom. If an armed robber is breaking into a home, and the occupant shoots the intruder, this will be documented as a gun-related homicide (which is technically correct). But is that actually a bad outcome? If the dead robber is to be tallied with the victims of “gun violence,” this requires the statistician to explain to the victims how their action was wrong, and to explain this without knowing the details of the particular case, or having any personal connection to the weight of that decision. The number-cruncher pays no price if the robber beats the home-owner to death with a crow-bar.

All of this is equally true in the other direction. A man can read all the John Lott papers and books, and keep his house and family fortified in a fortress bristling with defenses against potential dangers, only to discover that his son had found a loose pistol and had been playing with it in his room, when he accidentally. Here too, the statistician talking about how guns make everyone safer pays no price for being wrong. He isn’t even wrong. The statistics are just generalized.

Statistical analysis is, by its very nature, the removal of explanatory context surrounding events, turning complex interactions into fungible bits of data. This makes them more easily communicable, and makes them sound more objective. But it separates them from the more complex reality. You can’t generalize these things into patterns that will tell you what’s “better overall” for society, as though “better” was something quantifiable.

Not only do statistics not accurately describe reality; they also don’t capture the depth of feeling and value at work here.

Even when statistical arguments aren’t just being naive or dishonest, they aren’t taking the question seriously enough.

No one who saw a family member murdered, or found a sibling that had shot themselves in the midst of depression, or lost a child in a horrific mass-shooting, is pushing for stronger gun regulation based on data.

On the other hand, no one who cares about guns based on ideals is persuaded by numerical risk or anecdotes of tragedy. The second amendment was won with bloodshed. There are some things worth dying for, and many of our ancestors fertilized the ground with their bodies to give us the power to make these decisions for ourselves, to defend ourselves.

To throw away their sacrifice out of squeamishness would be cowardly.

To take the subject seriously,

Imagine your child, caught in the crossfire of a gang turf-war.

Imagine getting the phone call, hearing how one of your best friends growing up had been struggling personally, and no one knew it. They found him still holding his Glock, next to what was left of his mouth.

Imagine you’re apprehended on the street by an addict with a weapon.

Would you still support gun ownership, if the question was personal? After all, these are the experiences and realities that inevitably come with guns.

Logical points don’t matter here. No one cares that gun control isn’t that effective in keeping weapons out of the hands of criminals. No one cares that knives are also dangerous while they’re burying a young nephew who was in the wrong place around a careless family member and a negligent discharge. No one cares that hanging is also lethal after finding out their father shot himself after the divorce.

If you want to defend gun ownership, you have to defend it facing down the worst dimensions of guns — the horror, the uncertainty, the violence, and the tragedy. You have to bite that bullet, and say “yes, I accept that, and own it.”

This book is my attempt to do this.

The gun question is not a question about safety. It’s a question about what kind of people we want to be.

Perhaps we can build systems that refine our society. We can pass more laws that govern our behavior, which guide is in our decision-making. These systems might make us safer, healthier, even wealthier… not right away, of course. There will be bumps and failures along the way. But over time, as our tools improve, our health and safety will no longer be in the hands of fallible, untrustworthy people, but in systems guided by reason towards a better way of life.

Such a path will probably require us to give up our guns, because guns empower us as fallible individuals — often at our lowest and most emotional, least trustworthy moments.

Maybe that is a good thing. We can transcend the worst of humanity, the violence and animosity that seems to be embodied in the machine we call the “gun.”

On the other hand, we can continue more or less as we are, and always have been: as mechanically-inclined monkeys. We will repeat the tragic suffering of our ancestors… but at least we’ll keep our toys!

The thesis of this book is that for all the promise of the first option, the second option is better. We are shooting hominids. Ranged weapons are a part of our identity, and even a source of our sense of agency in our experience of the world. The sense that we have some kind of choice in what sort of life we are going to live — rather than blithely accepting whatever comes as a matter of “fate” — is in some sense a result of our skill in hitting things from a distance.

The connection between shooting and agency becomes clearer when we look at the role of bows and guns in history, in England and in America especially. The particularly English notion of self-governance appears to have emerged with the power of the longbow. But beyond civics, our very language points to a connection between shooting and morality. Ranged weapons — the “guns” of yore — guide our ideas of morality itself, as something distinct from virtue.

All the violent horrors that come with guns are simply technical enhancements of human nature itself. The path of systems that protect us from the foibles and fallibility of other people and their guns is a path toward safety only because it is a path of mediation and separation from other people. The reality is that humans are dangerous. We are still monkeys: monkeys with guns. That’s pretty dangerous.

But for all the danger and risk that can come with living with gun-monkeys, a life separated from other people isn’t worth living.

And if shooting truly is a part of who we are, the gun question might not only be about connection with other people. It might challenge our connection with ourselves, and with our own nature. We risk alienating ourselves.

But if we embrace that risk of living with gun-monkeys, and try to manage it personally, we might find ourselves better monkeys.