Tuesday, December 11, 2012

Every Picture Tells A Story, Part Two (Repost)

The gun prohibitionists are already saying that blood will run in the streets when the Illinois General Assembly crafts some form of concealed carry legislation in accordance with today's ruling by Judge Richard Posner of the 7th Circuit Court of Appeals.
“As the dissenting opinion points out, the two judges who threw out Illinois' law did not take account of the danger to the public from stray bullets, and they ignored the Illinois legislature's determination that carrying weapons has been shown to escalate violence,” said Lee Goodman, an organizer with the Stop Concealed Carry Coalition.
If Lee had taken the time to examine the graphic below, he would see that there is no positive correlation between less-restrictive (legal) carry (shall issue and constitutional issue) and increased violent crime. Let's face it - Chicago already has concealed carry by thugs and criminals. The only ones prohibited from protecting themselves are honest, law-abiding citizens.

This was originally posted on Dec. 21, 2011.

In October, I posted a graphic developed by Rob Vance that showed the progress in the growth of firearms carry rights from 1986 through 2011 as a percentage of the U.S. population. In 1986, 90% of the U.S. population lived in states that severely restricted carry rights or had none at all. Today, over two-thirds of Americans live in states with either shall-issue carry or constitutional carry. The conclusion was that shall-issue is the new norm.

Recently, the FBI released its Uniform Crime Reports statistics. Rob generated a new graph plotting these violent crime rates against the growth of less-restrictive firearm carry rights over the period of 1986 through 2011.

The data used to generate this graph is available here in Google documents. As I said in the original post, unlike the gun prohibitionists, we publish our data for the world to see. The data used comes from three sources: the FBI Uniform Crime Reports, the U.S. Census, and www.gun-nuttery.com/rtc.php which tracked the changes in carry laws over time.

Rob had the following conclusions after examining the data:
Violent crime is a complex issue, but national data is clear that there is no positive correlation between liberalized concealed carry laws and increased violent crime.

The “blood in the streets” or “Wild West” scenarios just don’t play out. To be fair, state level data is highly variable, so some caution needs to be exercised in drawing conclusions.
Rob is very explicit that there is no positive correlation between violent crime and liberalized carry laws rather than a proven negative correlation. He feels that it will take more rigorous statistical analysis before this negative correlation could be said to be proven.

The results here are reinforced by the study done by Linoge at Walls of the City which compared violent crime rates and firearm ownership. In that study, Linoge did find a relatively strong negative (-0.605) correlation between violent crime and firearm ownership.
While violent crime did not go down as sharply as the progress in carry rights rose, one thing that needs to be kept in mind is that while shall-issue may be the new norm that right is not extended to all locations. Bars, restaurants, government buildings, post offices, and many parks are still off-limits to concealed carry. I think this will become the next battleground for concealed carry. Interestingly enough, restrictive states like California actually have fewer restrictions on where one may carry.

As an example of the battle for fewer restricted locations, one need look no further than here in North Carolina. This session of the General Assembly changed the law concerning concealed carry in state, municipal, and county parks. It is now allowed everywhere with the exception of a local opt-out for "recreational facilities". As might be expected, some locales are trying to push the limit on what is considered a recreation facility far beyond what was intended by the General Assembly.

Shall-issue may be the new norm but the fight will continue.

NOTE: If you click on the graphic, you can view it in its full original size.


  1. While it can be quite hard to prove causation, it's painfully obvious when there is no correlation. They really set themselves up for failure when they make such dire predictions.

  2. Pyro, you're right on point about that. When the gun-grabbers spout off about how liberalizing concealed carry in the remaining states where it is currently restricted will result in some set of horrific consequences (aka wild west/blood in the streets), we should demonstrate that the history proves them WRONG, and badly. It's about removing reasonable objections. Then of course for some folks, "guns are bad, mkay?" is all they need. We need to convince the undecided/uninvested. I hope this helps that process.

  3. One of the things that the press has tried very hard to downplay also is that in all cases in which there has been a mass shooting (e.g. Virginia Tech), it is in a place that the shooter knows in advance is restricted to handguns.

  4. For extensive (with controls) corroboration, see Handwerker, W. Penn (2010) Why Violence? Because the Irrationalities Induced by Choice Frames Make Perfect Evolutionary Sense. Available at SSRN: http://ssrn.com/abstract=1523850

  5. Let's try that again: The shooter knows in advance has a restriction on handguns.

  6. Nice article Doc. From the abstract "strength deters violence and weakness elicits it." Painfully obvious to the rational mind, and a painful fact to be avoided by the irrational.

  7. Jimpithecus, yes, we understand, the shooters picked places where they knew their targets would be unarmed.

    I think, for most gun opponents, it boils down to fear of their neighbors.

  8. You don't "prove" a correlation. You find that a correlation is statistically significant and, therefore, likely to be actually occurring. Getting a statistically significant and negative r value isn't that hard.

    More rigorous statistical analysis is necessary to investigate whether condition A is driving condition B, whether the reverse is true, or whether it's coincidental. But that's not correlation. You're moving into predictive and causal analyses which move beyond correlation.

    For example, is the fact that fewer people are getting lashed with buggy whips over the last 100 years a result of decreased production of buggy whips or some other factor? There's going to be significant correlation between the number of lashings and the changes in production. But that's not proof that we're safer because it's harder to get buggy whips.

  9. My favorite example of this is a British study of correlations of cancer rates near nuclear power plants, which they found ... including sites where a plant had only be proposed, not built....

  10. I think the important thing here is:

    1 - The argument against gun rights is that it is a danger to the public.
    2 - For that to a valid argument, there MUST be a causation relationship between gun carry and ownership, and violent crimes.
    3 - Causation requires a positive correlation (as well as other proofs).
    4 - Repeated datasets show either no correlation or a NEGATIVE correlation.

    Therefore, the argument than guns cause crime CANNOT be true.
    Suppose there were a proposition that the moon causes things to get warmer, and elaborate explanations for how that was true.
    Objective measurement of night time temperatures show that they are colder than day time temperatures. Stop there.
    The elaborate explanations are irrelevant. The data shows a negative correlation between night and temperature. No further review of the elaborate theories is required. The hypothesis is show to be untrue.

  11. Nicely put Alan. And right on point. Thank you.

  12. I'm hard-pressed to find a practical reason that gun carry/ownership could be correlated to violent crime other than guns have a sort of hypnotic effect on their owners, sort of like a voodoo totem of evil that slowly steals souls. Would an anti-gun person chime in on this? I'm asking this earnestly.

  13. > Therefore, the argument than guns cause crime CANNOT be true.

    I sincerely believe that guns cannot cause crimes, any more than rabbits cause dogs, but an inverse correlation doesn't disprove causality any more than a positive correlation proves it. Maybe crime would have dropped *more*.

    What would be closer to proof would be comparing the crime rates on a state-by-state basis for states that legalized concealed- or open-carry, normalized for the date when the legalization occurred. If there is (as I expect) a drop in crime, or at least a failure to rise, timed appropriately, then you'll have your proof.

  14. This is no surprise, at all! Good job!

  15. >>What would be closer to proof would be comparing the crime rates on a state-by-state basis for states that legalized concealed- or open-carry, normalized for the date when the legalization occurred. If there is (as I expect) a drop in crime, or at least a failure to rise, timed appropriately, then you'll have your proof.<<

    The datasets are out there. Where states have liberalized gun laws, and rates of decrease accelerate faster than the nationwide downward trend, and faster than the pre-change downward trend.

    This is, I think, the crux of the Constitutional argument. If there is statistically either no correlation or a *negative* correlation, then the "public safety" justification of a compelling public interest CANNOT be true.
    The first test of any infringement of an right under "government interest" is that the infringement *must* accomplish its objective. If the objective cannot be accomplished, the infringement cannot be justified.

    So, with causation disproven via negative or neutral correlation, under *any* standard of review, the justification for infringement of a Constitutionally protected right cannot stand up.

  16. ALAN quote:>Therefore, the argument than guns cause crime CANNOT be true.<

    Correct on face value. A restatement is required. A device {gun, car, knife...} does not cause crime. People do. Attributing crime or any other events to objects is the work of weak thinkers that also believe in magic and avoid critical thinking.

    Cause and effect:

    So The presence of a significant deterrent {cop, alarm, owner with gun} crime is less likely. Why?
    Unless your insane, dying is a bad thing and a career shortening event and criminals maybe stupid but not usually insane (or drugged out).

    The statistical test is to compare those states that have tight gun control to those that do not.
    Do the curves match or or they different. I suspect the trend rates are very different with
    those "controlled" states vs more liberal carry states.


  17. @Alan,

    Sorry but that is wrong. All you have to consider is that there may be another, or a set of other effects having a stronger causation than the one you are studying. Do states with lower gun control also have something else happening, like higher awareness / education, or more serious penaltiesfor felons? In which case gun ownership may not be sole-handedly affecting the numbers.

    Also to be considered besides decrease in numbers would be the seriousness of the felonies. Does the burglar shoot first if the house owner wakes up because the latter is more likely to be armed? In which case felonies could be fewer, but more deadly, or more likely to happen in a given situation (if the guy makes it inside, you're in more danger).

    1. if there is another factor or set of factors I would absolutely love to hear about them. The graph here shows what has been studied and arguing on the basis of the theoretical is irrational. Over all, accounting for all reported cases and circumstances represented here, there is an easily interpreted negative correlation.

  18. This graph has a very simple purpose. It is not a model, nor does it have any of the many elements needed to build one. It is "descriptive statistics" only. Which is why the caveats stated up front. The purpose behind it is to illustrate that the predictions of mayhem made by the Brady's and their ilk did not come true (in a macro sense) as the fraction of the US population which had legal access to armed self defense increased significantly. Better statisticians than I will be needed to get much deeper than that. I do expect that if such a comprehensive model existed which had statisical validity in predicting public policy choices with respect to crime it would be well known even to non-specialists like myself. Even in the field which I do have graduate work (business) input output economic models have proven notorious for their inability to accurately predict the future for decision makers. In spite of the admitted limits of the analysis (and they are many), I stand behind the fundamental proposition illustrated by the work.

  19. Crime rises and falls over time, and gun restrictions rise and fall. I would like to see the results of some smart modeller taking into account a broader range of gun laws and comparing them with the crime rate. The 1960s and 70s brought forth a wave of gun control laws, and we know crime rose after that through to the 1990s. In the 90s guns controls were loosened, and crime has fallen. Maybe coincidence, maybe not.

  20. Note that one thing you must correct for is demographics: the more young men there are, the more crime.

  21. What Brady, MAIG, et al keep saying is "more guns on the street [and they falsely equate this to more CCWs] means more crime". This graphic does not address that claim. What is says simply is: "crime is going down as more people are living in more CCW liberal states". What we really need is crime rate (per 100,000)versus CCWs (per 100,000).

  22. "Interestingly enough, restrictive states like California actually have fewer restrictions on where one may carry."

    If you limit concealed carry licenses to the politically connected, you don't want to piss them off by restricting where they can carry.