Wednesday, January 12, 2011

Important Data Trends the NRA Doesn't Want You To Know

One silver lining to the mass shooting tragedy in Arizona has been a return, however brief, to a public discourse on gun control, and maybe some legislative action.  It's sad, frankly, that it takes the death of six people, including a federal judge and nine-year-old girl, and the wounding of 14 others, including a congresswoman, by some gun-nut loony in order to get us to this talking point, but at least we're here.  So let's talk.

There has certainly been a lot of conversations, too.  Discussions on the ease of getting guns despite having serious mental health issues, extended magazines that used to be illegal and should be again, the need for better background checks for gun purchases, and Arizona's lax gun laws.  As usual, the NRA has been mum on the whole tragedy.  No surprise.  Gun deaths are an inconvenience when you are trying to convince the public that more guns and less gun regulation is good.

A friend of mine shared an interesting article.  It's not really about the massacre in Arizona, but rather looks at a state-by-state comparison of gun regulation versus deaths, and ranking the states accordingly.  "20 Deadliest States" the headline reads.  How does your state rank?  Go to the link and find out.

But my friend went on to say he didn't quite trust it.  "The Wyoming data is a bit disingenuous," he wrote, "There are only about 700,000 people in the state, and they largely live rural lifestyles, even in the cities."  I thought this was an interesting comment, so I decided to crunch the numbers, and then I compared them to state populations and the density of population in those states.  I even graphed it out.  What I found was very interesting.

By now you've noticed the graphs.  Don't run away!  I promise they're painless!  I'll try not to immerse you in statistics and numbers.  Promise.  There's no test.   

Gun control advocates like myself are very aware of the alarming data that comes from studying shooting statistics.  We are also very aware that the pro-gun lobby wants you to ignore them.  It's okay, though, I know you probably aren't interested in seeing columns of numbers and all those horrifying statistics.  But here are some results that are very hard to ignore.

The first graph is a "bubble" graph.  Each bubble represents one of the 50 states (sorry, I couldn't figure out how to get Microsoft Excel to label each bubble separately) and plots it against a gun law rating that was assigned to that state by the Legal Community Against Violence and the number of deaths / 100,000 population.  This data is from the article I mentioned above.  That rating from LCAV is based on the number and type of gun laws in that state, such as background checks, mental health checks, concealed carry regulation, etc.  The higher the rating, the stricter the gun laws for that state.  Using the number of deaths per 100,000 people is a way of normalizing states for population differences (a value of 12, for instance, would mean that if you took 100,000 people at random, 12 of them likely would be shot in their lifetime), based on available shooting data.  I then created a bubble with a third set: population density.  The larger the bubble, the more densely-populated the state (number of people per square mile).  I then had the program draw a trendline, which illustrates any trends.

(correction:  the X axis should read "Gun-Related Deaths / 100,000".  There may be a few bubbles hidden by others.  Sorry, but I wasn't able to figure out how to add the state names, but you can click on the links to find individual state data.)

As you can see, there is a very clear trend:  The more strict the gun regulations in a state, the fewer people die.  The less strict the gun regulations in a state, the more people die.  THIS is the data that the NRA and pro-gun people don't want you to see.  It goes completely against their argument that having less firearms regulation makes for a safer society.  This is why they try to steer their arguments away from questions of community safety and deaths and start puffing up over constitutional rights to arm themselves.

The other thing you can see in the bubble graph goes to the point that my friend made.  But it's the opposite of what he and I expected.  What it shows, based on the sizes of the bubbles, is that those states that have the highest population densities actually have fewer shooting deaths per capita!  States like New Jersey, New York, and California, where people are more crammed together, have fewer deaths.  Conversely, those states that are less densely-populated and are more permissive with gun regulations, like Arizona and Alaska, have the highest per capita gun deaths.  It seems clear to me that better regulation outweighs the potential for violence that population density may introduce.

Oregon, where I am, falls right in the middle of the trendline, with 10 deaths per 100,000 and a rating of 30 out of 50, and we are the 12th smallest for population density.

The second graph is taken from LCAV data.  It directly addresses the pro-gun argument that more legal gun ownership = safer society.  Once again, their argument is hollow.  The exact opposite is true.  As you see by the trendline, states with a higher percentage of legal gun ownership have more per capita gun deaths.  Once again, Oregon falls right in the middle of the trendline, with gun ownership of about 39% and a little over 10 gun deaths per 100,000 people.  How does your state rank?  Click on the "data" link above.

I haven't scared you away with numbers yet, right?  Good.  Now, what are you going to do with this information?  Here's what you can do:  every time you hear pro-gun people state those same, tired, hollow arguments, send them here.  Let them see the numbers for themselves.  See how quickly they try to change the conversation away from issues of public safety!  And when they start hiding behind the Constitution and its second amendment, ask them what the document's main goal is:  protecting the welfare of our people, or mindless devotion to possession of deadly weapons?

UPDATE (1/19/11):  Here is another link with data that goes very well with this data, including maps of gun death rates by state and information on variables (such as poverty or strictness of gun laws) that have correlative associations:



  1. Thank you for sharing this very illuminating information. I will use the graphs freely when lobbying in the halls of Congress and my state Capitol.

  2. And we know the answer is:

    mindless devotion to possession of deadly weapons

    Followed by hollow slogans about freedom,liberty, and gun ownership.

  3. Wonderful, japete! Certainly. I'll send you the Excel spreadsheet.

  4. I mentioned this post in a comment at:

    The gun crowd ignored it.

    Maybe you have to figure out how to label the States.

  5. Thanks for forwarding the link, Laci.

  6. Can you please show me how the "rating" was determined? I can't find a hard number on their site, thanks!

  7. Sure, Andrew. I assume you mean the LCAV Gun Law Rating in the top graph. Here is a link to their rating system description:

  8. I have a few questions. On the first chart:

    - What data did you use for "deaths?" Was it the FBI statistics? CDC? Someone else? Are they just violent deaths, murders, murders involving firearms, suicides, all firearms related deaths, some combination, or all deaths total regardless of cause? Did you use one year of data or multiple years (taking an average)?

    -- If you just selected "gun deaths," what is the relationship to total violent deaths? What about overall violent crimes? Do we find that gun deaths go down, but knife deaths, rapes, and aggravated assaults go up?

    - How did you establish causality? You state your findings in a manner that suggests that there is a causal link. Can you say that "less gun control caused more deaths?" For example, another interpretation of the correlation is that in places with higher populations, politicans need to be seen as "doing something" about crime and thus pass gun control laws.

    - How much of violent death variation is explained by different levels of gun ownership? What is the statistical confidence level you in your correlation?

    - What other variables did you control for which might explain the variation in gun deaths?

    Regarding the second graph:

    - Why did you use firearms ownership rates from 2002 and gun deaths from a different year?

    - Are you aware of the potential biases in the 2002 BRFSS? For example, 30 states reported data that was weighted heavily to certain locales (maybe cities, maybe rural areas, who knows -- we just know it isn't representative). Did you consider that people from states with strict gun control laws might be less likely to report owning firearms? Why did you select the 2002 BRFSS as your source for firearms ownership data?

    - Why are there not 50 dots? Are some occluded because they overlap? Or were some states excluded? If so, why?

    - How did you establish causation for your thesis of "more guns causes more violent gun death?" There may be a correlation but it could be to other factors. For example, in a state with high levels of violent crime, more citizens might purchase firearms for self defense. In this case, the high crime rate is causing the increased gun ownership rate.

    Forgive my skepticism. A large number of peer-reviewed studies have found that there is support for the thesis "more guns = less violent crime" and "more guns have no effect on violent crime rates." I am only aware of one peer reviewed study that claims "more guns cause more violent crime" and it has serious methodological errors. So your findings might superficially look startling but I think if you start digging into your data set it may be harder to support some of the conclusions you have drawn.

  9. Now subtract suicides and justifiable homicides from your "gun death" numbers and we'll talk.

    Roughly that is 10,000 deaths in the US each year (and only 60% of those actually use firearms as the weapon). However I will use 10,000 murders as a baseline simply to make my point.

    Over a population of 300 million that is less than 3.5 murders per 100,000 annually. Since none of your bubbles shows a rate close to 3.5 you have artificially inflated your numbers with suicides, justifiable homicide, etc.

    Redo your numbers, with a little intellectual honesty, and you might be surprised to find that the trendline is much flatter than you wanted.

  10. I would have to confess that I'm very confused. First of all, you just include gun deaths. I may be wrong about this, but shouldn't you be using murder statistics instead? If a policeman were to shoot someone robbing a store, that would be a gun death, but not a murder. For that matter, if a robber stabbed and killed a gun store owner, then that would count as a murder, but not a gun death.

    This is important, because the NRA and their ilk claim that legal guns push down the overall murder rate.

    Second, I find your graphs confusing--especially the first one, which seems to cloud issues by cramming in as much data as possible. The first graph is especially confusing, for several reasons:

    -- By leaving off the names of the States, I can't see what's going on. Yes, I know that's Excel's fault, but it's still a problem. (Are you sure that Excel can't do this? I seem to recall that Excel is very good at graph generation.)

    -- Is it "volume of the spheres" or "area of the circles"? If it's the first, then a lot of data will be obscured, because we're only getting an area representation of a sphere, which wouldn't be good.

    -- What year of statistics are you using, and how do graphs compare to other years? I'd like to know, because low-population states are more susceptible to fluctuations in murder rates. To take as an extreme example, a small town of 1.500 people could have 0 deaths per 100,000 for decades, but two murders could cause that to jump to 115 per 100,000 for just one year, and then the sleepy town could go for another two or three decades with a death rate of 0 per 100,000.

    Ah, heck! I'm sure that my questions could be answered if I could look at the data myself. Do you have a link to it, or have it available for download? Thanks!


    Here is a link showing violent crime rates, not just gun crimes, by country. You will notice that several countries with very restrictive gun laws have much higher violent crime rates as compared to the United States.

  12. I find it fascinating that Illinois has a rating of 8 while Chicago has a rating of 15.5 according to 2007 data. Kind of takes away from your whole point, doesn't it?

    And to contrast your data, D.C. has a rating of 31.2 and they have the least permissive, besides Chicago, of course.

  13. OK, you cite the Daily Mail article, which is a conservative paper. Unlike the US, British papers make no bones about their being biased.

    "The Tories said Labour had presided over a decade of spiralling violence."

    Secondly, the date on the article is from 2009--meaning that was a Labour Government.

    The issue is that there was an allegation that the crime statistics were underreported. Also, there were allegations that the crimes were placed in lower categories than they were.

    If that is the case, then US crime statistics are also doubtful since there have been similar allegations in Miami and Philadelphia. Rape is also underreported.

    If you are alleging that crime is non-existant in areas where guns are present, then that is as false, or even more false than the allegations that gun control prevents crime.

    People who advocate gun control do not allege that it is a panacea for crime, but that it will reduce otherwise preventable deaths.

    On the other hand, "pro-gun" forces allege that the presence of firearms prevents crime.

    A claim which was proven false this past weekend in Tucson,AZ.

  14. Jadegold posted this at MikeB's

    Jim, in comments, raises the old gunloon canard that the UK is a far more violent place than the US, South Africa, and most EU countries.

    He cites this article. In a nutshell, The Daily Mail argues that the UK has a violent crime rate of 2,034 per 100,000 citizens, the US has 466 per 100K and South Africa has 1,677 per 100K.

    Now, if I were a gunloon, I'd attack the messenger. I might point out The Daily Mail was big champion of Adolf Hitler, in particular, and fascism, in general. I might point out The Daily Mail has lost a number of libel suits to various people, including Elton John, Diana Rigg, and Kate Winslet to mention a few.

    But I won't.

    Instead, I"ll point out The Daily Mail fails to note crime statistics vary from nation to nation, both in the way they're categorized and the way they're tabulated. For instance, playing one's stereo too loud is counted as a violent crime in the UK--in most other countries, it may not even rate a misdemeanor charge. Similarly, the UK has what is called an affray which consists of 2 or more people fighting/arguing in a public place--this, too, is categorized as a violent crime. In the US, such an instance might rate a misdemeanor.

    It's very easy to see The Daily Mail piece doesn't pass the smell test. We know homide rates in, say South Africa are about 10 times higher than the US and nearly 40 times higher than the UK.

  15. Strange.

    Do you include suicides among 'gun deaths'? Does removing suicides from the data change the trend, or make the correlation weaker?

    If we re-run the data with cities instead of states, does the correlation change?

  16. My state, WI, is #36 on that list. Let me tell everyone about gun laws in WI. ...there aren't any. Other than no concealed carry, (which we will be getting soon in one form or other), there are no permits, no registration, and a 2-day handgun waiting period. There is, of course, the standard background check and forms.

    What this means is I, a law-abiding citizen, can buy a semi-auto version of an AK47, with a 30-round, heck, let's make it a 40-round mag, and be out the store's door with it in 15 minutes.

    Oh, and I can open carry without a permit, too. All of that freedom, and the state is still way down at #36. Remove the armpit called Milwaukee, and WI would probably be #46.

    Washington D.C.'s murder rate has gone down for the two years its gun ban has been repealed. National crime is down while citizens buy more guns than ever and carry more than ever.

    What does all of this tell us? It tells us that guns themselves aren't the problem.

    Also remember how many deaths from guns are via suicide and via gang wars. If the U.S. put as much effort into stopping gangs as it did trying to ban guns, the country would be a much nicer place. Look at Mexico. Strict gun laws, but drug gangs run wild with military-grade firearms and explosives. Some might argue legalizing at least certain drugs would fix that problem easily, but that's a different debate, I s'pose.

    As for Arizona, anti-gun-rights people often say to let the cops handle it. Where were they this time!? Citizens took down the shooter. And then some talk about the armed citizen on the scene who Didn't pull out his gun, yet no one is talking about the cop in NJ(?) this week who was shot by his fellow officers. If people think cops have some magical training and superpowered, quick thinking no one else has, sorry, but they don't.

  17. Wait a minute...did you just try to relate deaths by all causes (ranging from sloppy driving to cancer) to total gun ownership?

    What is the postulated cause/effect linkage?

    How does gun ownership affect cancer rates? Deaths by automobile accident? Deaths by cardiac arrest?

    I'd love to know. Because if you think there is a connection, then the world of medicine is missing out on something that you know, and they don't.

  18. Hi, Karrde. No, those are deaths due to gunshots, not all deaths. My X axis should be named the same as the other graph ("gun deaths / 100,000").

    However, "gun deaths", as defined by the source data, I think does include homicides, accidents, and suicides, as well as justified shootings by citizens and law enforcement. Some posters have taken exception to that, but I say it's valid -- even in the case of justified shootings, since most justified shootings are due to the perpetrator being armed as well.

  19. Oh, as to whether the trend changes if you remove suicides, I cannot say, since I haven't run the numbers. Nor have I seen any city-by-city comparisons. If you find such data, please share.

  20. He's not a "gun-nut Loony"

    He's a Loony who happened to use a gun to hurt and kill people. He could have just as effectively used a car to do what he did.

    I understand that you are trying to create dialog here but your credibility suffers when you use terms to describe someone as a "gun-nut Loony" when "Loony" would have been a sufficient and accurate description.

    You are unnecessarilly stirring the pot with the "gun-nut" descriptor.

    Have a great day!

  21. If you find such data, please share.

    Go check out John Lott's website. He has made his data set -- which includes county level data -- available for download so you can use it however you like, including to look at your research question.

    As a research question pretty closely related to your graphs, Lott has found zero peer reviewed studies that substantiate the hypothesis of "shall issue permits (i.e. more guns in public places) increases violent crime." Of course, obviously Lott is biased on this issue as he has a stake in defending his research, so if you can point me to any peer reviewed studies I'd be interested to read them:

    I kind of doubt that many such studies exist though.

  22. The #1 ranked LCAV state is California, with 19.5% household firearm ownership and an 8.98 gun death rate.

    The #48 ranked LCAV state is Vermont, with 45.5% household firearm ownership and an 8.38 gun death rate.

    How can that be?

  23. Appreciate seeing the visuals. Wish gun control laws were seen more a common sense, rather than stifling. Jasmine

  24. That last chart is quite possibly the worst linear fit to a scatter plot I've ever seen. Just because you can draw a line through scatter data doesn't mean that you've identified a meaningful trend or linear system. You've weighted the entire response by the lowest 5 data points which represent 10% total per state (not capita) population. Imagine removing the lowest five and tell me how your system responds then. Ideally you'd have more or less the same fit, perhaps with greater uncertainty in the fit (something you've conveniently left out of your calculus).

    If you did this in a proper science or engineering publication you'd get hung high by reviewers. I get so tired of bias by both sides of any argument, be it guns, race, politics, or whatever.

  25. @ Unknown: Actually, there are 6 dots there, each one representing a state. Whether you like those points or not, they are valid, or would you prefer to just drop out six states of the union? And they aren't "weighted" in any way; The data isn't treated any differently for those six states than for the others. You can't just discount the data because it makes a line that you don't like.

    I would also add: I deal with statistical analysis every day as part of my job, including publication of data, so you can't get away with an accusation that I am somehow getting away with "bias" or am unfamiliar with statistical significance. Wishful thinking won't make the data turn out differently for you.