"walls of the city" logo conceptualized by Oleg Volk and executed by Linoge. Logo is © "walls of the city".

graphics matter, year the second, part two

By now, even irregular readers of my weblog should be more than aware of my yearly posts documenting how the hypothesis “more guns = more deaths” is demonstrably false over recent American history, but today we are going to go a slightly different direction.

In a recent post, Joe Huffman positively shredded this asinine quote from Paul Helmke:

More guns mean more gun violence, sometimes intentional or sometimes accidental.

… and I pointed out, just for the record (since Paul was referring to overall crime and not just fatalities), that there is effectively no correlation between the raw numbers of firearms and firearm-related fatalities, and a negative correlation between firearm ownership rates and firearm-related fatality rates, but the above blockquote got me to thinking (always a bad thing) – what about crime, rather than just fatalities?

To begin with, we are going to have to define the problem, which Paul Helmke, in his usual weasel-worded fashion, did a very poor job of – what is “gun violence”? For the purposes of this post, we will arbitrarily take that phrase to mean, “any recorded crime committed wherein the aggressor was armed with a firearm” – crimes committed with firearms, or CCwF for short. Alright, so the question is, “Do firearms, or firearm ownership, cause CCwF?” since “means” in that context is just another word for “causes”.

Great, so where are we going to get our information? As with previous years, the American population numbers will come straight from the Census Bureau; the number of firearms in civilian hands will be constructed off a combination of the Small Arms Survey of 2003 and the BATFE Annual Firearms Manufacturers and Export Report; and the number of shall-issue carry states was cribbed from this helpful graphic at Radical Gun Nuttery. However, this year we will discard the CDC’s WISQARS system (since we are not interested in fatalities), and instead turn to the best clearinghouse of information pertaining to crimes in America that I am aware of: the Federal Bureau of Investigation’s Uniform Crime Reports.

Based on that source, we will be totaling murders, assaults, and robberies committed with firearms* between the years of 1995 and 2009**, and comparing those numbers against the number of firearms in civilian hands, and the overall population of the country… kind of like this:

As with every time before, it is important to understand that I adjusted the various orders of magnitude in order to get everything to fit on the same graph, and while the two images may look similar, it is also important to understand that this graph is a full order of magnitude higher than the previous “graphics matter” exercises. In short, the line “Crimes Committed with Firearms” is left entirely unadjusted; however, both the “American Population” and “Number of Firearms” lines must be multiplied by 1,000 in order to get their “real” numbers. Additionally, the rate lines are presented in magnitudes of “100,000,000 firearms”, “100,000,000 people”, and “100,000 people”, as explained by the graph’s legend and annotations. These changes do not affect the accuracy of the information presented (especially since I kept all the digits of each relevant data point, despite moving those numbers’ decimal points around) – the important thing on this graph is trending, not specific numbers (although once you multiply by the appropriate order of magnitude, the numbers are still correct).

So what does the above graph tell us about the hypothesis “more guns = more ‘gun violence’”? Pretty obviously false. But how false?

As we all know, “correlation does not imply causation“, but the rest of that phrase continues as, “… but causation requires correlation,” so is there any correlation at all (and thus the possibility of causation) between the number of firearms in civilian circulation and the number of crimes committed with firearms?

Well, unlike last time, I will not bore you with the full barrage of equations, charts, and whatever else, though I will provide this reminder: “‘r’-values can range from -1 to +1, with +1 meaning that all data points lie perfectly on a line, with Y increasing as X increases, and -1 means the same with Y decreasing and X increasing. 0, logically enough, means no correlation at all.”

Looking at the raw numbers of firearms and crimes, and solving for linear (Pearson) correlation both by hand and by way of Excel’s onboard linear regression tools, we find that their ‘r’-value is -0.35744, which indicates that there is a weak, negative correlation.

Looking at the firearm ownership and crime rates (both on a “per 100,000,000 people” scale), and again solving by hand and by Excel, we find that their ‘r’ value is -0.59499, which indicates a strong (though not very strong), negative correlation.

If, however, you are genuinely interested in those equations, charts, explanations, and numbers, and actually want to delve into the nuts and bolts of this analysis, you are certainly free to dig through the entire, updated spreadsheet available here – unlike anti-rights nuts, I have absolutely no problems maintaining my integrity and honesty by showing my work and making it available for constructive criticism, error-checking, and examination. Likewise, do remember that all root numbers on that spreadsheet come from such authorities as the FBI, the Census Bureau, and the BATFE, and all calculated numbers are generated from only the information provided by those sources.

So what are our conclusions?

1. The hypothesis “more guns = more ‘gun violence’” is demonstrably false over the past 15 years of American history, based off the information available to us. The number of firearms in civilian circulation has been steadily increasing over that time period, and the number of crimes committed with firearms has not been equivalently increasing – in fact, it has been fluctuating in a fashion that might be described as “wildly”. However, again, since there seems to be some confusion on the concept, proving “more guns = more ‘gun violence’” to be false does not necessarily prove “more guns = less ‘gun violence’” to be true. Doing so would require accounting for far more variables than I did, and involve far more interesting math than I employed.

2. When comparing raw numbers, there is a weak, negative correlation between the number of firearms in America, and the number of crimes committed with firearms. This negative correlation completely precludes any possibility firearms causing crimes committed with firearms.

3. When comparing rates, there is a strong, negative correlation between the number of firearms per person in America, and the number of crimes committed with firearms per person. This negative correlation also completely precludes any possibility of firearm ownership causing crimes committed with firearms.

Is that not interesting?

One final conclusion is that Paul Helmke lied (again), this time by way of omission – by not doing the relatively simplistic work (so easy I could do it) of crunching numbers that are freely and publicly available to anyone with an internet connection, Paul expressed a falsehood as truth… which is kind of sad, considering that verifying your statements before you say them is only common sense. I do not know if, like a disturbing number of anti-rights nuts, he simply does not know the difference between fact and fiction, or if he simply did not care to make the distinction – in either case, Paul lied, and in a blatant, easily-disprovable fashion at that.

Oh, and what the hell is “accidental” violence?

As always, the spreadsheet documenting all of the number-crunching I performed is freely available, and you are certainly welcome to use the above graphic however you like – I would, of course, appreciate a linkback were you to repost it anywhere.

* – For whatever reason, the FBI only keeps track of weapons employed for those three crimes – murder, assault, and robbery. I am fairly certain that rapes, motor vehicle thefts, home invasions, and whatnot else are committed with firearms as well, but the FBI could include those instances under the appropriate headings of “murder” or “assault”.

** – Those are the only years that the UCR documents the specific weapons used for crimes. If anyone knows of any sources going farther back than that, I would certainly appreciate the information – the more years we have, the more-accurate this analysis can become.

23 comments to graphics matter, year the second, part two

  • How dare you use research, data, and facts to make a point. Where is your use of emotion and “won’t someone think of the children”?

    What the hell kind of lobbiest are you? :-)

  • Beautiful work, Linoge! I wonder how long it will take for an anti-rights nut to refute it out of hand, and maybe make an outlandish claim, like you made up your numbers, or that the FBI is secretly beings controlled by the NRA.

    These posts are awesome both in how clearly they support our claims, but also the open and honest way you created it, which of course stands in stark contrast with how the other side conducts themselves.

  • Dr. Feelgood

    For whatever reason, the FBI only keeps track of weapons employed for those three crimes – murder, assault, and robbery. I am fairly certain that rapes, motor vehicle thefts, home invasions, and whatnot else are committed with firearms as well, but the FBI could include those instances under the appropriate headings of “murder” or “assault”.
    The UCR employs a hierarchy rule that could mask some crimes involving firearms. If an incident is reported in which a person is assaulted or robbed with a firearm and forcibly raped then the UCR scores only the forcible rape; and as you noted, the FBI does not track weapons for forcible rape. The UCR Handbook details this Hierarchy Rule (page 10) and its exceptions.

    It would be hard to know how this affects your analysis. Perhaps you could make a reasonable assumption that forcible rapes are committed with weapons at rates similar to robbery, assault, and/or homicide. Be generous to the other side in this assumption and see what happens.

  • @ hsoi: The unpaid kind. If someone is willing to pay me, I might be willing to parrot whatever talking points they are paying me to parrot :).

    @ Weer’d Beard: Thanks! … And, yeah, I can already imagine the anti-rights nuts lining up to decry my math (because, you know, calculating rates gives you all manner of wiggle room…), or my sources, or whatever the hell else comes to their benighted minds…

    And I cannot honestly say as though I care :). The facts are what they are, and if those facts blow their precious little beliefs out of the water like so many paper canoes, so be it.

    @ Dr. Feelgood: In other words, RTFM? :)

    That explains it, though… If I get bored, I might go back and apply the percentage-of-crimes-committed-with-firearms for assaults to rapes, and factor that into the overall data as well, but, honestly, given that the ratio of crimes seems to remain unchanged throughout the years (while the total number of crimes can vary whackily), I am not convinced it would change the end result – the slopes would be more-or-less constant, which is what linear regression calculations are most concerned with.

  • Looks like they already have, and I think you know who and where.

    Don’t even bother reading it, its pretty embarrassing…

    The other side really isn’t healthy or normal people…

  • Dr. Feelgood

    Yeah, that thought occurred to me, too. There might be a slightly steeper slope if you use the firearms rates for robbery, since they’re substantially higher than for assault. That masking effect is a major flaw in the FBI’s crime data, in my opinion. If they’re going to place forcible rape hierarchically above assault and robbery then they ought to track weapons-use for rape. They do give homicide figures for weapons-use where homicides occurred in conjunction with a rape. Those numbers are surprisingly low; and, for 2009 at least, there were zero homicides concurrent with rape that were attributed to firearms.

    Anyway, the robbery-by-firearms rate shouldn’t change the overall interpretation, but it gives you mildly more defensible data. Not that the GFWs are open to things like reason and facts.

  • [...] jobBob H on motivation… to get a better-paying jobAMB on helping a heroDr. Feelgood on graphics matter, year the second, part twohelping a hero | walls of the city on hat trickweambulance on motivation… to get a [...]

  • Playing With Numbers | Where Angels Fear To Tread

    [...] [...]

  • @ Weer’d Beard: Believe you me, I am not going to waste my time trying to comprehend what they think they are saying… As Joe Huffman amusingly quipped yesterday, I could have a more-meaningful conversation with a variety of small pets than I could with those individuals.

    And you will note, after all these years, that not a one of them has ever bothered to pen up their legitimate, mathematical proofs that I am wrong, and present them… Instead, they rely entirely on hysteria, emotion, and “studies” carefully crafted to appease those who are paying for them.

    As Kevin Baker also recently said, I do not give a damn about studies – I care about raw data. And there it is.

    @ Dr. Feelgood: I concur. Burying data like that is only going to lead to confusion, and if they are going to hide robberies and assaults that were committed with firearms behind the larger category of rape, then they should break out which rapes were committed with which tools as well. Bleh. When I get bored again, I will sit down and project out the rape-committed-with-the-assistance-of-a-firearm numbers based off the percentage of robberies-committed-with-the-assistance-of-a-firearm, but I am going to maintain that and the above as separate data sets – we are venturing into the realm of speculative deduction, and I want to make that both clear and disjoint :).

  • MIke Menkus

    You might be able to find firearm crime going back in time from the Crime Victimization Survey. Here a summary report that might be helpful … check out page 9

    Here’s a link to older Crime Victimization Reports:

    An idea for your chart ….. change the number of states that allow carry to POPULATION in states that allow carry. Its a measure that will blend better with your other stats.

  • mike Menkus

    Have you seen this quote ….. I thought it was interesting that DC keeps two sets of books about crime.

    The D.C. police department’s statistics and those from internal documents, as well as those kept by the FBI, have routinely differed. Two weeks ago, the FBI released numbers that showed violent crime in D.C. went up 7 percent in the first six months of 2010 compared with the same period last year. The city’s statistics, however, show violent crime as being down 7 percent. Police Chief Cathy Lanier told
    The Examiner that the FBI’s report “is not a good measure of District crime” because D.C. police go by the city’s code to determine the D.C. crime rate.

    Read more at the Washington Examiner:

    In my studies, I’ve noticed that crime stats are corrupted for various reasons in gun control areas. The UK changes the definition of crimes, Australia just does not collect data, and most don’t have a “real” baseline available prior to the implementation of gun control.

  • I had not seen the Crime Victimization Surveys… that will get me at least another two years’ worth of data, once I dig through them – thanks for pointing them out.

    As for the carry statistic, the problem is that the whole population of the state is not eligible to carry, neither will all of those eligible actually carry; and given that almost none of the states actually have specific numbers of carry permits going back that far, I am stuck with numbers of states rather than people.

    Regarding the blatant corruption of data by those interested in hiding it, as I think I said way back in the first of this series of posts, I am quite interested in skewing the raw numbers to the anti-rights side of the spectrum – hence my choice to include suicides in the previous graphs regarding firearm fatalities. In this case, if certain municipalities underreport their crimes (and we all know they do), then that is just another way I am helping the anti-rights nuts.

    Why am I doing that? Because it proves that even with counting suicides, or even with the diminished crime rates, there is still no correlation between firearm-related fatalities / firearm-related crimes and actual firearm ownership. In short, I give them just enough rope to hang themselves :).

    That, and I am too damned lazy to dig through all the reports and suss out who is lying and who is not :).

  • Facts Don’t Matter | Where Angels Fear To Tread

    [...] [...]

  • [...] Graphic Matter – by Linoge [...]

  • Great article, well articulated and very well researched. You have done a great service to free America and your time and effort is much appreciated Linoge.

  • Thanks, Zeus! Feel free to repost it wherever you like, and spread the good word – facts like these need public exposure to best destroy the hysterical, emotion-driven nonsense of anti-rights nuts.

  • Hm. In doing my homework, there appears to be a small glitch with the Crime Victimization Surveys – they operate on interviewing the victims, and if the victims are dead, they cannot really interview them. Guess those will not be quite as useful as I had thought…

  • Sorry I’m so late on this. Linked!
    Great job as usual!

  • [...] commentsLinoge on graphics matter, year the second, part twoLinoge on No resolutions, but I do have a goalLinoge on in reviewLinoge on coming to a mall near [...]

  • [...] Red light cameras strike a special nerve with me.  The arguments for their use are much like the arguments for gun control, it makes everyone safer.  The actual numbers from the field though show the complete opposite however again, just like gun control. [...]

  • [...] When Linoge published his three graphical analysis of guns and crime in America Here, Here and Here. He was not “Diverting attention AWAY from guns” but simply expanding the field to [...]

web analytics

View My Stats