46 thoughts on “Your gun control ideas won’t work. This one will.”

1. Thanks Zach. Great analysis of a difficult topic. Taking your call to action to heart

   Reply
   1. Thanks for reading and for your comments, Bob! It’s great to hear from you :)

      Reply
2. Very interesting analysis. Thanks for the hard work of pulling it together. The numbers describe the coincidence between income inequality and homicide, but I wonder how you (rhetorical you) would explain any underlying causality. Do the poor (over-simple label, I know) envy the rich (ditto) and kill them, or do the poor envy the rich and take out their anger by killing their fellow poor, or do the poor try to steal from the rich and the rich shoot them? What’s really going on underneath the connection itself? That might affect what happens as you try to fix income inequality, depending on how you try to fix it. If you’ve got tension become income classes and you try to fix the inequality by simply redistributing from the rich to the poor, you might exacerbate the tensions (and maybe the willingness to commit violence) by making the rich resent the redistribution (and the poor who benefit from it). And would redistributive programs have any effect on the willingness of any group to commit violence? On the other hand, if you try to fix the inequality by promoting general economic growth (assuming that a rising tide lifts all (or most) boats) that might not happen. And would such a growth approach have any effect on the willingness of any group to commit violence? Or a host of other possible scenarios I’m probably overlooking. Seems like there are a number of interesting variables that need to be better understood, so the proposed remedies improve the situation rather than making things worse.

   Reply
   1. Thanks for reading and for your insightful comments! These are some questions that I’d love to address in the near future. My understanding of homicide in general is that most crimes have a single victim, most victims know the perpetrator, and in most cases the victim and perpetrator are of the same race. I believe that there are likely many underlying causes wrapped up in the Gini coefficient: Better access to mental health care, a better social safety net for women who want to leave abusive relationships, a more libertarian drug policy that makes the underground trade unprofitable, etc. At the highest level I believe that people will defect against a society that they feel has abandoned them when they have nothing left to lose, and that defection often involves violent crime.

      Reply
3. Where does the firearm ownership rate come from? Averaged, those state figures don’t add up to the ~100 guns / 100 residents that are quoted in most current sources (and Wikipedia). Granted, the state numbers are from 2001, but if the disparity between then and now is so great, surely we need more up-to-date numbers to draw any conclusions?

   I’m asking mainly because your table of comparison of US vs Canada has them both at nearly identical firearm ownership levels – 31 vs 32 – with 32 for US, I’m assuming, being the state aggregate? – but there’s just no way there are as many guns per capita in Canada *today* than there are in US (and I say this as an American gun owner who lived in Canada before). And it seems that this particular number being so close is partly responsible for why you get such a low contribution coefficient for it.

   Can you plug in the 100 firearm ownership rate for US into your model instead of 32, and see what it gives then?

   Reply
   1. Thanks for reading, and great question. The firearm ownership data are at the household level and come from a 2001 survey (the most recent available at the time) that I used as the source for my 2013 article https://zachmortensen.net/2013/01/10/gun-control-four-must-see-graphs/

      Here is the link to the dataset: http://www.washingtonpost.com/wp-srv/health/interactives/guns/ownership.html

      The 100 guns/100 residents number is a top-down estimate at the national that isn’t useful for analysis of variance in violent crime at the state level. It comes from various estimates of ~330M guns (no one knows for sure how many actually exist) / ~321M residents. I know of no reliable dataset that shows how those supposed ~330M guns break down by state.

      As far as I know the BRFSS data linked above is still the most recent, most comprehensive survey of firearm ownership that exists. If you find a better one with data at the state level, I will gladly incorporate it and re-run my models. Or if you know where I can find a couple million dollars, I’ll design and run a new study myself!

      Unrelated: Your username made me nostalgic for hooking BIOS interrupts in my assembly language programming days. Thanks for that :)

      Reply
      1. I meant, use the model that you’ve trained on 2001 data (on the assumption that it’s still valid), but feed it the 100 aggregated number for the country specifically for the purpose of comparing it to Canada. I’m just curious how much effect it’ll have – I would still expect Gini to dominate by far, and if it does, it would strengthen your argument even further, but it’s interesting to know either way.

         As far as total number – all these numbers are going to be very rough estimates, simply because there are no gun or sale registries in US. For what it’s worth, the oft-cited 300 million guns figure is considered a low-end ballpark estimate within the gun community itself, with 400+ million being more likely, just based on the yearly number of NICS checks for the past few years, and eyeball estimate of how many of those are for new vs used guns. If I remember correctly, the 300 million figure was actually computed in the same manner, just earlier.

      2. Thanks for clarifying, and what you ask makes sense. I’ll post a snapshot of it here in the comment thread. Stand by…

      3. So it turns out that I more or less did this scenario already, it’s the one I summarized in this quote in the original post:

         “Getting the same effect from gun control alone, according to the model, would require disarming twice as many US households as actually own guns, an impossibility.”

         In other words, a reduction of 64/100 households would put the effect from disarmament on par with that of a Canadian Gini index + US GDP. You’re asking for a reduction from 100 down to 31, or 69/100 households. The magnitude of the effect from that would be slightly greater than that of the Gini.

         But again, this is garbage in/garbage out with the current model weights. You’re taking a model that is trained on a household firearm ownership rate and feeding it a per-capita number, which the model interprets as 100/100 households owning firearms, which is overstating the status quo by a factor of ~3x. So yes, if 100% of US households owned a gun, the model would predict a reduction of ~18,000 homicides per year by disarming 69/100 of them. But it would also predict >35,000 homicides per year as the starting point, which is nowhere near what we have today.

         Here’s a recent study, less than a year old, from a reputable research organization that also found 30% of US households own a gun, in 2017. Sample size was 1,500 so it’s not rich enough for state-level data like BRFSS, but it’s enough to validate that the 2001 numbers are still useful: http://assets.pewresearch.org/wp-content/uploads/sites/3/2017/06/06151541/Guns-Report-FOR-WEBSITE-PDF-6-21.pdf

         I’ve found other sources that estimate Canada’s household gun ownership rate as low as 25%. So disarming to Canadian levels would be going from the range of 30-32% of households down to 25-31% of households. Some uncertainty in the ranges of course, but given the small factor weight it’s not going to make a big difference.

      4. What changes if you don’t use a raw rate for ownership? The number of guns in the US is often deceptive because an increasingly small number of people own an evermore larger number of firearms. It stands to reason that it shouldn’t matter for measuring crime or homicides after a certain threshold. That is, what difference does someone owning 20 guns rather than 2-3 make? In any case, I’m with you on the income inequality. There is more than enough evidence about that being the source of the vast majority of our society’s problems.

      5. Great question, and thanks for reading. The ownership rate in the model is in fact a household rate, because that’s how the most reliable data set (the CDC’s Behavioral Risk Factor Surveillance System, BRFSS) gathered its data. I agree that the total number of guns likely matters less than the number of households/individuals who have access to at least one gun, so we are fortunate to have the data aggregated this way.

4. Ohh, I see now. That’s the part I misunderstood – I thought that number *is* the per capita rate. You might want to highlight the “household” part, because, while it’s mentioned in the article, it’s only once – and thereafter it just refers to “firearm ownership rate”. For someone doing cursory checking, it ends up looking weird, because if you put “firearm ownership rate” into, say, google, you get plenty of tables – but they have that 100 number in them.

   (I agree that tracking households with guns, rather than number of guns per capita makes more sense.)

   Reply
5. This is dumb. This is really, really dumb.

   It uses averages for all shootings. It doesn’t pull out mass shootings or those committed with specific types of firearms. Statistically, mass shootings are outliers. Statisticians are often trained to exclude outliers, but that doesn’t work when minority data that do not fit into the statistical model involve bodies, including the death of children.

   Looking at the averages this guy uses, he explains away gun violence by correlating it to “diversity” and economic disadvantage. How many mass shooters are not white males – aka “diverse” – and how many are from below the poverty line? Considering that we know the answer, how do solutions that address non-correlative factors solve the mass shooting problem, which cannot be lumped into shootings overall?

   You need to go beyond just stats to a qualitative understanding of the mass shooter issue, including the firearms used. What is the social benefit of persons who do not plan on executing mass shootings owning weapons of that power, which are impractical for hunting and other non-murder activities? If these weapons are to be used for their intended purpose (military-level combat resulting in numerous enemy casualties), what civilians should be allowed to commit mass shootings, and how does society benefit?

   If you want to go into gun violence overall, though, let’s return to the argument that violence correlates with “diversity” and economic status. Arguing against gun control from a position of white male privilege, then, does so at a cost to persons with socioeconomic disadvantages.

   You’ve got an article that bases the entirety of its argument on statistical correlation to broad cultural trends. However, you don’t find the word “regression” in there for regression analysis, which is the standard means by which statisticians control for other mitigating factors that may impact an issue. Cool, bro. Cool, cool.

   The opening rhetoric is garbage. Typical structure of multiple lines: “Yes, other countries with fewer guns have fewer gun deaths.” He does not address any of the data points he introduces in those lines in his actual statistical analysis.

   Two important numbers are not included: 36, and 25.2. The United States has 36 gun homicides per one million residents. For the 23 most economically advanced countries in the world, second place (Canada) is 5 per one million. That makes our gun homicide rate 25.2 times the average for those other 22 nations. We simply do not differ enough from those other countries culturally or economically to account for a difference of that power – and again, that difference is counted in bodies.

   The primary statistical difference between us and other, similarly developed nations is the number of guns and ease of access. Nobody with even a quantum of statistical literacy could write or read this article and believe that this analysis negates that simple fact.

   https://www.cnn.com/…/amer…/us-gun-statistics/index.html

   Reply
   1. > You need to go beyond just stats to a qualitative understanding of the mass shooter issue, including the firearms used. What is the social benefit of persons who do not plan on executing mass shootings owning weapons of that power, which are impractical for hunting and other non-murder activities? If these weapons are to be used for their intended purpose (military-level combat resulting in numerous enemy casualties), what civilians should be allowed to commit mass shootings, and how does society benefit?

      If you want to argue from a position of demanding more accuracy, you really need to do your part and get more familiar with types of guns that you’re targeting.

      For starters, you might want to look at the kind of weapon that Anders Breivik used.

      Reply
      1. Anders Breivik. So the weapon that a person who did a mass shooting in another country – and is thus outside of this data set – used is relevant to this matter how, exactly? Would you like to note how many times AR-15s have been used in American mass shootings in recent years? I doubt it would help your argument.

         The author of this has written two gun control pieces: One published right after Sandy Hook, another right after the recent Florida High School shooting. He is timing these to argue against any form of gun control right after children have been killed. Yet he tried to fold in mass shootings with shootings overall to come to an illogical, unsupported conclusion that no form of gun control could mitigate this country’s issues. Considering the timing of both pieces, obfuscating issues that are unique to mass shootings is deliberately obtuse.

         We have supposed regression here, except not nearly enough data points to call it that with a straight face. Nothing on magazine size or type of weapon used. Nothing attempting to correlate data to other predictive factors, such as histories of domestic violence or mental health holds for persons being a harm to himself or others. If other factors had been included in the regression, the casual dismissal of limited gun control measures as a means of preventing gun violence would be absolutely ridiculous. The author chose data points to fit his narrative rather than using the maximum amount of relevant data to discover a narrative, and the end result is utterly irresponsible.

         BS’ing research to “Well, you know…” your way over children being murdered is pure sleaze.

      2. Brian, are you judging my sincerity because I chose to publish my work at a time when the public discourse on this issue is occurring? Would I have been more believable if my timing was better?

      3. Sincerity doesn’t matter when analyzing data and methodology. Your work is valid, or it’s not. As a PhD-level quantitative researcher who passes peer review a few times a year and peer reviews stats-based articles for a few journals, your research is utterly lacking in validity. You flat out did not include enough variables for proper regression analysis. You need far, far more data than what you cherry picked to be able to confidently draw a conclusion about the root cause of broad societal issues. The concept that you included enough variables to confidently say that we have seven times as many gun homicides as Canada, and 25 times the annual per capita number as the 22 other most wealthy nations in the world, can be explained away through income inequality would be laughable if not for the bodies. Your analysis is invalid without any regard to your intention.

         As for your intention, however, your timing is suspect and your decision to cherry pick data is even more questionable when it comes to your intention. You are sealioning with data.

         https://www.urbandictionary.com/define.php?term=Sealioning

      4. > Anders Breivik. So the weapon that a person who did a mass shooting in another country – and is thus outside of this data set – used is relevant to this matter how, exactly?

         The reason why it’s relevant is because it clearly shows that you don’t need an AR-15 to pile the bodies up – any semi-auto rifle will do just fine. Which, in turn, means that any assault weapon ban is going to have an efficiency of about zero.

         And unless you have reasons to believe that guns operate differently within US borders as they do outside of them, the country where it happened is entirely irrelevant.

         What does this have to do with this study? Nothing, really, I was just responding to your assertions.

   2. Brian, thanks for reading. The analysis was in fact a multiple linear regression across the variables discussed, and if you read carefully, you’ll see that I offered a link to the file containing the model and the underlying data in case you want to critique the actual work rather than my writing.

      Also, you may be interested to read the following by a researcher at UCLA who has arrived at similar conclusions: https://luskin.ucla.edu/connection-poverty-inequality-firearm-violence/

      Reply
      1. On the UCLA / Mark Kaplan study, you’re kidding, right? You are dismissing most forms of gun control. Quotes from you:

         “To my friends who find themselves in the middle, neither passionately for nor against gun control: Your instincts are right, gun control will not solve the problem.”

         [People who refuse to engage in discussions of economic inequality] “are complicit in allowing the statists to frame violence exclusively as a gun issue.”

         In short, guns are not the problem. You find correlation between the number of guns and number of gun deaths, but you move to quickly dismiss this. And “statists” helps to show your bias, which I believe shaped your manipulation and cherry picking of data in your highly flawed and incomplete regression analysis.

         Meanwhile, quotes from Kaplan:

         “‘The reason for putting this up is because quite often the discussion about guns is sanitized.’ Kaplan explained, ‘We often treat the issue of gun violence as an abstraction but when you talk to people who are working in emergency departments, when you talk to coroners and medical centers, this is what people tell you. This is the net effect of guns.'”

         “A relationship exists between the high rate of guns and gun ownership and the number of homicides, suicides and injuries. Kaplan said that work is needed ‘on the ground’ in America’s cities, such as limiting access to guns that would ‘go a long way to reducing some of the fatalities that we’re experiencing both in the homicide and suicide.'”

         “‘The more guns, the more lethal the assault. When guns are absent, people are more likely to survive an assault,’ said Kaplan. ‘If we could just tamp down the levels of gun ownership, that might – might – be the first step in trying to reduce the rate of gun violence.'”

         —

         Any reading of Kaplan that dismisses gun control, and thus does the opposite of what you suggest, goes completely against his argumentation. Even if you find some parallels between your data and his, your reading is way off base.

6. Hi! I really like your post. I was wondering, did you look separately at homicide committed by police? I’m having a hard time figuring out how to use the NVDRS from the CDC (I’m not a statistician) but my understanding from some other articles is about 10% of homicides are done by police in the line of duty. Do you have any idea what impact that would have on your model, assuming the scenario where civilian guns are removed but police remain armed?

   Reply
7. I’m part of a group called The Liberal Gun Club. Would you mind us re-blogging this article on our website? We’ve just put out a few things so timing-wise it could be a few days or weeks, but I think your article’s great and, assuming the leadership agrees, it represents a good analysis and speaks to the root-cause efforts we think are necessary to solve gun violence.

   Reply
   1. Thanks for reading! And yes, you are welcome to re-post with attribution and a link to the original.

      Reply
8. Interesting but this does not address the phenomenon of mass shootings in the USA, I think it was an error to analyze homicides individually. Could you explain why there are more mass shootings in the USA than in a country like Venezuela which has the biggest homicide rate in the world? I can see how a lower Gini coefficient could reduce the number of homicides but can it also reduce mass shootings?

   Reply
   1. Thanks for reading and for your comments. The data include all mass shootings, I chose to not analyze mass shootings as a separate phenomenon because they are relatively rare compared to homicide in general, and result in ~100 of the ~17,000 homicide deaths per year in the United States.

      Also, the definition of “mass shooting” is somewhat variable.

      Here is a link with some historical data and discussion of the various definitions of a mass shooting: https://en.wikipedia.org/wiki/Mass_shootings_in_the_United_States

      Reply
      1. While I’ll admit to the naïveté of the question, for what reason is the hypothesis of removing all guns from the equation not discussed?

         I can make comprehensive arguments for alternative choice of firearms that would create entirely different scenarios vis-á-vis mass shootings. Why is that option never considered?

      2. Thanks for reading and for your question.

         That option is certainly on the table, at least theoretically. The statistical modeling exercise is an attempt to understand what might happen if we were to go there.

         I’ll grant that if you could, in an instant, remove all firearms there would be no mass shootings. There would be no gun homicides. There would be no suicides by gun. Most people seem to imagine these desirable outcomes and ask, “Why not?”

         But what would happen to homicides overall? To suicides? To robbery and other violent crimes? The effects here are not as intuitive, therefore the use of statistics to study the underlying factors and possibilities.

         And pragmatically speaking, what if this hypothetical war on guns turns out like the failed war on drugs, or prohibition? Making something illegal, as difficult as it would be to pass legislation banning and confiscating all guns, seems historically easy compared to getting the bad guys to obey the law so that we achieve the desired effects.

         So no option is off the table. I just don’t want to see society waste time and political capital pursuing anything that won’t work.

   2. If mass shootings have social causes – which is strongly implied by the fact that not only they predominantly happen in one country (US), but even in that country they’re limited mostly to specific socioeconomic and cultural groups, it would be reasonable to expect Gini to play an important role, since it reflects many of those socioeconomic factors.

      The question with respect to Venezuela is perfectly valid, but conversely, there’s also an equally valid question about Czechia – they don’t ban assault weapons or hi-cap mags, they have shall-issue concealed carry… and they don’t have mass shootings. Why?

      Reply
9. Great article, Zack. Me and a few other pro-gun people have been talking about how ending the War on Drugs and trying to lower barriers to getting jobs/involvement in the economy would be a better solution than gun control, so it’s nice to have numbers that back that up.

   That said, I feel like the reason no one really talks about it is two-fold:
   1. Both parties in the US have radically different ideas on how to accomplish those goals and refuse to work together/compromise, so nothing gets done. (For example, Republicans cutting regulations to make it cheaper to start/maintain a business vs Democrats taxing businesses higher to pay for social programs.)
   2. The political will and financial costs make them complete non-starters, due to politicians fearing backlash from the voters (many of whom have been conditioned to accept only one view of the problem) and their monetary sponsors.

   I feel like the reason so much of the debate revolves around guns is because they’re a politically convenient scapegoat for Democrats, whose platform planks have arguably led to/exacerbated a lot of the existing problems, and a political symbol for Republicans, since they represent a connection to an ideal of self-reliance. It’s a lot easier for both sides to just argue over what are inanimate hunks of metal than admit that a lot of their ideological talking points and/or past legislation are what got us into this mess. And unfortunately, I don’t think that’s going to change anytime soon, since the groups that got us into this mess are the only ones who can craft laws to fix the problems.

   Reply
10. I would be really interested to see this model applied to Australia before and after their NFA ban. It seems like this is the go to argument in favor of gun control, so it’d be interesting to see if the model accurately predicts the changes in homicide rates as a result.

    Reply
    1. That’s a great idea. I’ve seen data that “gun deaths” decreased, but that finding would be so obvious as to be trivial. The question I’d be interested to ask would be about homicides in general. I’ll look for a dataset.

       Thanks for reading!

       Reply
11. I don’t know how you can talk about factors influencing gun violence and leave out gender. It invalidates your entire point, when an overwhelming number of mass shooters are men, and one of the biggest indicators of whether someone will kill someone else with a gun is whether or not they have a domestic abuse record. It strikes me as odd, to put it nicely (plain racist to put it bluntly), that you would look at racial diversity and ignore gender. Frankly, if we put a blanket ban on men owning guns, we would greatly reduce the number of gun related deaths, and maybe we should just try it for a year while all the dudes are forced to take a class in gender and empathy studies 101 taught by a very nice green tea sipping middle aged lesbian professor.

    Also, your title for this article is “Your gun control ideas won’t work, but this one will” and then fail to mention what your damn idea is. Is it being nicer when when we talk about gun? Is it making sure to remember that “diversity” is the problem?

    This entire article screams one things to me: you’re the problem and instead of taking a breath and admitting that, you’ll twist any data points possible to blame someone else.

    Finally, it seems all your data is looking at violent crime, the definition of violent crime has nothing to do with guns. You can commit a violent crime with a knife, a bomb, your bare hands, your pet gorilla. According to the FBI, “violent crime is composed of four offenses: murder and nonnegligent manslaughter, forcible rape, robbery, and aggravated assault. Violent crimes are defined in the UCR Program as those offenses which involve force or threat of force.” So, if you came to my house and said “I’m going to rape you and if you don’t submit, I will find your mother and kill her” that is a violent crime, even though there is no gun involved. Taking stats from that and applying it to gun violence is all kinds of messed up. You briefly note this, but it seems you have absolutely no data on gun violence specifically, so I’m unclear as to how you can draw a conclusion on anything related to gun control without having gun crime specific data….

    I agree with your point on income inequality, and we should definitely be looking to fix that, not just for a reduction in crime, but all kinds of reasons. But, fixing income inequality takes decades, generations, of work in every aspect of society. Meanwhile, wouldn’t it be nice if we didn’t also give people guns to commit crimes?

    Reply
    1. “This entire article screams one things to me: you’re the problem and instead of taking a breath and admitting that, you’ll twist any data points possible to blame someone else.”

       That’s an interesting position to take on this issue, shooting the messenger like that. No pun intended.

       Which “data points” are you suggesting I twisted? I’ve made the dataset available, please be specific.

       Did you not read my call to action? I’m literally calling out *all* of us — pro gun-control people, anti gun-control people, and those in the middle — to reframe the problem and engage differently.

       I’m not opposed to you or anyone else advocating for gun control. I’m offering a statistical model that suggests that those policy ideas will not solve the problem they purport to address. And that same model suggests that addressing income inequality (the “This one will.” in the title) will have a 100x greater effect on the homicide vs. that of gun control.

       The model and the inferences I’ve drawn from it may not be convenient for your cause, in which case you seem to favor questioning the model rather than your own beliefs. That’s understandable. But at some point you need to ask yourself: “What would it take for me to change what I choose to believe?”

       In the meantime, I’m continuing to do the same. Thanks for reading and taking the time to comment.

       Reply
       1. The fact that you take data from violent crimes and not gun crimes for one. That is a HUGE misleading piece of information to draw from. Yeah, no effing shit if we reduce income inequality the gun related crime will reduce because ALL crime will reduce. I don’t know if you purposely excluded gun violence related data, but you did, and it invalidates your entire argument.

          Gun control advocates aren’t naive enough to think that any one law will solve all our problems (except maybe a law the outlaws corporate political donations), but several gun control position would make a significant dent, including:
          – forcing states and local law enforcement to share their information with federal agencies
          – actually enforcing laws that exist
          – allowing the ATF to use computers
          – outright ban on anyone owning any kind of firearm with a domestic violence charge
          – some sort of buy back program with significant incentives

          Tackling income inequality is a huge project, and it will take decades to properly address. In the meantime, we can not hand people guns willy nilly. You’re basically saying we shouldn’t have a free lunch program for low income school children because the real problem is income inequality. Yes, it is, but also, give the kid a sandwich because he is hungry right now.

          Reframing the problem? Sadly, no, your idea won’t work. For the reasons I mentioned above, and because it’s basically passing the nuclear football instead of attempting a solution. Historically, here’s the thing about societal issues: the only thing that works is just fucking doing something. No amount of mental reframing gets people on board, and eventually things come to a head, and part of the population is just never going to be happy about it. Sometimes, we have to try and we have to fail too. We’ve been having this discussion for decades now, reframing it here and there. I know you think your idea is novel, but it’s not. It’s a trite regurgitating of everything I’ve heard for the last 20 years. Meet in the middle, work together. And nothing changed. Any reasonable legislation that was proposed was somehow lobbied out of existence or effectiveness. Pick any major political social issue since this country came into existence, I will show you historical evidence of people “reframing” the problem for a good 30 years. Reframing the problem is rearranging deck chairs on the Titanic. Pick literally any gun control idea and pass it, enforce it and see if it helps. When it fails, we can talk.

          To quote Eddie Izzard, “guns don’t kill people, people kill people. But the gun helps.”

       2. I didn’t exclude gun crime, Dee. Every homicide in the country, knives and guns and cars and fists and baseball bats and pointy sticks and kids killed by AR-15s in school shootings, they’re all there. I regret that you didn’t understand that before you left such a lengthy comment on the premise that I had done otherwise.

       3. “And that same model suggests that addressing income inequality (the “This one will.” in the title) will have a 100x greater effect on the homicide vs. that of gun control.”

          Really? Is that a statistical measure? How did you, mathematically, come to that conclusion? When you’re saying you’re numbers aren’t bunk, you don’t get to make stuff up. The level of the impact you suggest is a guess, plain and simple.

          Meanwhile, factual number: 25 times as many gun-related homicides per capita in the U.S. as the average for our 22 most economically developed peer nations. We’re at 7 times that of our second-place peer. For an article attempting to smarm-splain gun policy to folks [right after a bunch of high schoolers were murdered, BTDubs], you flat out have not performed an analysis that explains away that magnitude of a difference.

          And let’s cut down past Dee’s rhetoric to ask a direct question: Why is racial diversity a meaningful variable, but gender is not? If we should look at your data and respond accordingly, why should targeting specific communities by race be acceptable but targeting by gender is beyond the pale?

       4. Brian, I applaud your engagement here.

          The 100x figure came from the sensitivity analysis described in the article. Based on your claimed PhD-level work I’m sure you’ve done at least one of these before. If you need help deciphering it, please let me know.

          Perhaps the root of some confusion on this issue lies in the choice of dependent variable. Many, including you, are talking about the connection between “gun deaths”/gun homicides and the number of guns. Yes, these two variables are strongly correlated, kind of like proximity to water is correlated with death by drowning. Reducing the number of guns or the number of households that own them will, of course, reduce the number of “gun deaths”.

          But that’s not the variable I’m studying. I’m looking at the overall homicide rate, which includes gun homicides (and not gun suicides, more on that later). Part of the reason for this is that I believe causal factors exist for homicide that have nothing to do with guns, and a near-infinite number of substitute deadly weapons exist for all but mass shootings (which account for only ~100 of ~17,000 homicides per year).

          There is broad variance in the homicide rate, the firearm ownership rate, and other potentially causal factors in the population of each state. I’m building a model to predict the overall homicide rate using these factors. I included racial diversity only because there is variance in that factor from one state to the next, and it turned out to be statistically significant. I could have included gender; perhaps you believe that there is a meaningful difference in the gender mix of the population of each state. But my non-PhD-level experience — and I grant that this has not been peer-reviewed — is that no meaningful difference exists in the ratio of men to women from one state to the next.

          Of course crimes of violence are committed disproportionately by men. Has it occurred to you that the economic equality represented by the Gini coefficient likely includes measures of “man control” as underlying factors? e.g. education and labor force participation by women, wage equality, free or subsidized childcare, cultural attitudes toward domestic abuse, etc. I’m not sure why you would choose to believe otherwise, or imagine that I somehow refuse to believe that gender equality is part of the equation.

          You’re obviously intelligent, and I’m not stupid. I’m happy to continue engaging with you if you can tone down your hubris a bit.

       5. I feel like you have a profound misunderstanding of data and statistics if you think you can make a solid conclusion about the gun debate by using data where gun violence isn’t specifically addressed…. It just doesn’t make any sense… It’s not a left or right thing, it’s a research thing. If I’m trying to answer the question of “what led to an allied victory on the eastern front of Europe during WWII,” I cannot make all my data points “overall allied victories in WWII.” I have to look specifically at victories on the eastern front, and what happened specifically in those battles. There’s nothing wrong with using your data per say, it can provide good context, but you’re drawing a larger conclusion about gun violence and what causes it WITHOUT LOOKING AT GUN VIOLENCE SPECIFICALLY. This is not rocket science. You’re excluding not just a major data set, but THE major data set. If you’re not drawing out those data points specifically, you might as well be excluding them, because you’re not going to be able to the necessary information from such a large data scope.

       6. Dee, I don’t know how much more clearly I can say this: The data that you seem to think I’m excluding are in fact included in my analysis. If you refuse to believe that, I can’t help you understand it.

          Studies that look only at “gun deaths” can’t help us understand what happens to homicide in general as we tighten gun regulation. Does the overall homicide rate decrease, increase, or stay the same? And if we prevent people from getting killed by guns, and the same number (or more) people are killed by other means — because there are non-gun factors that influence homicide and many, many substitute weapons exist for all but mass shootings, which are only ~100 out of ~17,000 homicides per year — then what have we accomplished?

          Studying the link between guns and gun deaths is a bit like studying the link between water and drowning. I accept that guns are related to gun deaths, and that relationship is not what I have chosen to study. You don’t seem to understand this.

          I have decades of experience doing this type of statistical work, and many clients who have benefited greatly from it; so there is no need for me to defend my competence in this domain. You are of course under no obligation to agree with my methodology or my conclusions; but it’s clear to me that you have failed to understand even my central research question, in addition to what it means to “exclude” data from a statistical analysis.

       7. They’re not complaining that you have excluded gun death. They’re complaining that you haven’t excluded everything else. Which makes no sense to me, but…

12. Interesting read from a Canadian who knows maybe 3-4 people that actually own a gun and those are hunting weapons. I have a few points that bothered me about the article but good on you for trying to come to a conclusion.

    I am not entirely sure of what your conclusion is about though and perhaps that is the biggest issue I have with the article. It will be used by Gun Control advocates to show that gun’s don’t contribute a lot to the overall homicide rate in the US. However, is that really the topic that is being discussed these days. The topic about Gun Control currently in the media is about reducing the violence being perpetrated by guns. So while your article clearly lays out that socio-economic conditions can reduce the homicide rates more than gun control I expect your article is going to be used out of context to say that Gun Control doesn’t really affect crimes committed with guns. But that is reader beware I guess.

    Onto some facts that I just have a slight issue with (but no solution to get different numbers).

    I know you have explained your rate of Canada’s household gun ownership vs US household gun ownership but I would have to think your numbers of 31 and 32 are not accurate. Of course I tried to find out more details but I was amazed at the lack of information on gun ownership for either country in the past almost 20 years. Your number for the US is based on a poll from 2001 when the US had 40 million people fewer (which is one Canada’s worth of fewer people). The poll only asked .07% of the population the question. Is that enough of a sampling to accept the number as accurate? I’m not a stats person but it seems awfully low to base a number on. The reports I found on Canada generally look at a 1 in 4 or 25% household ownership but those reports were even worse, dating back to 1994. And of course using numbers from 17+ years ago both from a homicide point of view and a gun ownership rate amongst your other contributing factors doesn’t necessarily mean it is valid today. 17+ years is a whole generation of of youth and individuals that have grown up in a completely different culture than last century.

    The website I found (http://www.pewresearch.org/fact-tank/2017/06/22/key-takeaways-on-americans-views-of-guns-and-gun-ownership/) suggests that 42% of those polled live in a household with a gun. This is likely in a different format that what your charts but interesting and more recent.

    Overall though, I think you also need to take into account the type of gun’s available. I would imagine most homicides and gun-violence activities are done with handguns and it would be interesting to see how the numbers flesh out without grouping all firearms as equal. In Canada, for instance, the report I read from the government suggests that 97% of the households that own a gun own a long-gun but only 12% of those households own a hand-gun. While the report above doesn’t give quite the same numbers they state that 66% of gun owners have more than one gun (doesn’t split out by type) but of the households with only one gun, 62% of those are handguns, 22% rifles and 16% shotguns. What I’m saying with that is that comparing Canada’s household numbers to the US’s numbers, even if your above numbers are correct, would not be accurate in determining effect given the large disparity in types of guns owned.

    Anyway, thanks for putting together this article as it can be used as a conversation piece regardless of opinions about the data or conclusions drawn.

    Reply
    1. I appreciate your insightful comments and the time you took to read the article and respond. I agree that there is too little data available on the number and types of weapons that exist in the public, and the numbers in every survey are likely under- rather than over-reported, i.e. more guns probably exist in more households than are willing to respond to any survey in the affirmative.

       I think you’re probably right about the distribution of long guns vs. handguns as well, i.e. the US likely has far more handguns per household than Canada does, and handguns account for a disproportionate share of the homicides. I think that long guns account for far less than 10% of gun homicides, but I’d need to confirm that with the data the FBI has published.

       I’m in the process of drafting a follow-up article to address the differences in the various measures being used in this debate:

       Homicides: Murder, manslaughter, etc. Non-accidental deaths at the hands of another by any means.

       Gun homicides: The subset of the above that occurred by means of a firearm.

       Mass shootings: The subset of gun homicides that involve 4+ victims (this definition varies considerably btw). These are rare by comparison, ~100 deaths out of ~17,000 total homicides per year are due to mass shootings.

       Suicides: Intentional, self-inflicted death by any means.

       Gun suicides: The subset of suicides that occurred by firearm.

       “Gun deaths”: This measure encompasses gun homicides (including mass shootings), gun suicides, and accidental firearm deaths. Many articles point to a strong correlation between the number of guns and gun deaths, which is about as obvious as saying that proximity to water is correlated with drowning.

       The problem is that we could reduce “gun deaths” to zero and, except for eliminating accidental deaths, do very little for society because reliable substitutes exist to commit suicide and homicide (with the possible exception of mass shooting events).

       Stay tuned!

       Reply
13. Pingback: Lies, damned lies, and gun statistics | Zach Mortensen
14. Thanks for the clear and thorough presentation, and thanks most of all for doing the dirty work. My background is in history and law, and lately i’ve nurtured a suspicion that orbits your analysis as a sort of corollary: that perceived economic inequality tends to drive elite demands for firearm restriction. I’m no statistician ( which is why I’m especially grateful for your work here),and i’m no paranoiac, but I submit that if more ppl take an honest look at work like yours, and at the history of gun regulations (and indeed of popular movements), many would perceive a rather frightening nexus between the wealth gap and violence–whether of the state or private persons. Thanks again.

    Reply
15. I’d say it’s ironic that you’re getting the most disdainful feedback on your analysis from the gun-control supporters posting here, but I’d be lying if I said I was surprised by it. Most gun control supporters have a very strong emotional attachment to the idea that the data do (and must) favor their position, and it can be jarring for them when things don’t quite turn out that way.

    One issue I would take with your hypothesis correlating homicide with income inequality is that homicide has been dropping steadily over the last couple of decades, but we are led to believe that incoming inequality is *increasing*. How would you propose to test this hypothesis, given the claim that economic disparity is (at least, we are told) getting worse?

    Much clearer is the lack of correlation between guns and homicides (which is the central point of your thesis, of course) as we have added near 100 million firearms (in the vast majority semi-automatic handguns and rifles) into the mix over the last couple of decades. There’s simply no way we can honestly say that an increase in firearms in a society leads inevitably and automatically to an increase in homicides (or even homicides by gun) – as difficult as that may be for many to accept.

    Reply
16. Since Ive noticed that very few individuals who have a license / permit to possess firearms make the evening news in a negative way, I would, in a more sane world, be in favor of requiring one to possess them. I would make it “shall issue” but require “reasonable” training. Not the “You cant get there from here hoops to jump through as in many jurisdictions, but enough law and survival techniques to keep from appearing on the evening news in a negative way.

    I would not require registration of individual firearms, as many times those who “are not coming for your guns” have lied and done just that subsequent to their “reasonable” gun-registration schemes.

    In a reasonable world, this would be ok. However, in the current America, where when someone misuses firearms, the screaming-heads demand unspecified “action” (They’d like a total gun ban, but just wont say it), this will remain impossible as too many innocent gun owners are, in fact…innocent

    Reply

Leave a Reply Cancel reply

Enter your comment here...

Fill in your details below or click an icon to log in:

Email (required) (Address never made public)

Name (required)

Website

You are commenting using your WordPress.com account. ( Log Out /  Change )

You are commenting using your Twitter account. ( Log Out /  Change )

You are commenting using your Facebook account. ( Log Out /  Change )

Cancel

Connecting to %s

Notify me of new comments via email.

Notify me of new posts via email.

Δ