Skip to main content

A front-page diary today by DarkSyde inexplicably touts this blog by Howard Nemerov, an Austin (MN) gun rights advocate, for employing "science and analysis" instead of rhetoric to arrive at his conclusions.

Interested, I randomly checked out one of Nemerov's recent posts, which happened to be about the relationship between suicide, gun ownership, and the unemployment rate.

What I found, in spite of DarkSyde's claim, was a lot of rhetoric and not much science or analysis.

In his blog, Nemerov purports to test two competing claims: one by the Wall Street Journal that rising unemployment causes more suicides, and one by the Brady Institute that increased gun ownership causes more suicides. "Who's right?" asks Nemerov (apparently unaware that both could be right -- or both could be wrong, for that matter.)

Nemerov first draws unemployment data from the Bureau of Labor Statistics, and suicide data from the CDC.

Nemerov presents us with the following graph:
Nemerov1

Notice anything odd here? Anything missing? The alleged purpose of Nemerov's entire post was to investigate whether unemployment or gun ownership was the cause of suicide. So where's the gun ownership data here?

Missing, that's where. Nemerov proceeds to do a little hand-waving, pointing to tiny goes-up-here, goes-down-there changes in the suicide and unemployment rates, and decides that the Brady Institute must be wrong, and the WSJ must be right. I mean, come on. He doesn't even try to do a simple correllation, folks. This, apparently, is what passes for "science and analysis" in some circles.

Well, that's just not good enough for me. So I decided to do what Nemerov didn't do: some real science and analysis. Using the same data, or better, that Nemerov used.

Here's the first oddity I noticed: Nemerov confines his analysis to the years 1990-2006, because his desired age range (16 and up) cannot be teased out of the CDC suicide data farther back than 1990. However, if he had changed his age range to 15 and up instead, he would have found that the data goes all the way back to 1981. Odd that one year would make such a difference, but it does, so that's the data I'm going with here.

Gun ownership rates have been part of the non-governmental Generalized Social Survey (GSS) since the early 1970's. Unfortunately, the GSS isn't conducted every year, but we do have every other year recently, and a little more than that in the 1980's. But that's enough to do statistics and draw correllations.

That leads us to this graph:
suicide
Suicide rates per 100,000. Unemployment rates in per cent. Gun ownership in percent * 0.2, for scalability.

A few of interesting things jump out. First of all, the gun ownership rate and the firearm suicide rate track very, very closely. Second, the unemployment rate and the suicide rate, while both going generally downward during this period, aren't tracking nearly as well. Third, the non-firearm suicide rate stays pretty steady and doesn't track with either unemployment or gun ownership at all.

We can formalize these eyeball guesses by running correlations across these variables, to see what variable is linked with what, and how strongly. When we do that, we find that gun ownership is correlated with the gun suicide rate in a highly statistically significant way (correllation coefficient r = .921, while the rule-of-thumb for significance is r > .7). Meanwhile the correlation between gun suicide and the unemployment rate is only .553, below the edge of statistical significance. Total suicides are correlated with unemployment a little better at .629, but that's still below statistical significance. But the correlation coefficient for total suicides and gun ownership also improves a bit too, to .932, so going with total suicides instead of firearm suicides doesn't buy Nemerov anything.

Having done that, let's try another exercise. Let's run a statistical model to try and use both unemployment and gun ownership to explain suicides. Here, let's use the total suicide rate, because that's a little better for the "unemployment explains it" crowd. Let's start by setting up an equation that might model the suicide rate, like this:

Suicide = (a * ownership) + (b * unemployment)

In other words, we're assuming that both gun ownership and unemployment contribute to the suicide rate, and we're trying to statistically determine how much of each goes into it. If we include all relevant causes when trying to model an effect, our model will improve and explain more of the variance. There's a standard algorithm that can do the stats for us (I use an implementation called Nonlin by Phillip H. Sherrod).

The results are interesting:

variable estimate error t prob(t)
a0.22140.0150014.780.0000
b-0.14980.1049-1.430.1751

The key thing to keep your eye on are the t statistic, and the prob(t) in the right hand columns. T tells us how far away from zero the variable is in relative terms, and prob(t) tells us the likelyhood that the variable really is zero. Here, the variable b, which is the one associated with unemployment, has a 17.5% chance of really being zero. In other words, we cannot statistically eliminate the null hypothesis for unemployment.

We can, however, eliminate the null hypothesis for gun ownership very easily. The prob(t) statistic is flat zero.

In other words, once we control for the effect of gun ownership, unemployment has no statisically significant effect on suicide rates. In fact, the best fit for variable b, the unemployment variable, is negative -- implying that unemployment might actually cause fewer suicides. One cannot make that claim, however, because it's still statistically insignificant regardless of what side of zero it's on.

The combined gun ownership-plus-unemployment model explains 80% of the variance in suicide rates. An alternate model using gun ownership alone explains 79% of the variance in suicide rates. So the difference is trivial, meaning adding unemployment doesn't improve the model much -- and it only improves the model at all because the residual correlation is negative, i.e., if we believe that unemployment causes less suicide.

It turns out that the Brady Institute is right, and the Wall Street Journal is (as usual) wrong.

And that's some science and analysis you can count on.

Geeky update:

It occurred to me after posting this that my simple model above was a bit too simple; I should have included a free variable to allow the model to have non-zero suicides if the gun ownership rate and unemployment were both zero. So instead of the above equation, an improved version looks like this:

Suicide = (a * ownership) + (b * unemployment) + c

... where c is the free variable. When you do things this way, the overall fit of the model improves a bit, now explaining 84% of the suicide rate. The same model without unemployment explains 83% of the suicide rate, again showing that unemployment makes no difference. The variable statistics do change somewhat, although the conclusions are the same. Here's the new variable stats:

variable estimate error t prob(t)
a0.17830.02467.260.0000
b-0.11060.0960-1.150.2697
c1.6390.78232.100.0563

The case for unemployment becomes even worse, with variable b now far beyond significance. Free variable c is sort of on the edge of significance, at about 6%. Variable a, for gun ownership, is still highly significant, though not quite as high as earlier.

Originally posted to The Numerate Historian on Sat Dec 12, 2009 at 01:08 PM PST.

EMAIL TO A FRIEND X
Your Email has been sent.
You must add at least one tag to this diary before publishing it.

Add keywords that describe this diary. Separate multiple keywords with commas.
Tagging tips - Search For Tags - Browse For Tags

?

More Tagging tips:

A tag is a way to search for this diary. If someone is searching for "Barack Obama," is this a diary they'd be trying to find?

Use a person's full name, without any title. Senator Obama may become President Obama, and Michelle Obama might run for office.

If your diary covers an election or elected official, use election tags, which are generally the state abbreviation followed by the office. CA-01 is the first district House seat. CA-Sen covers both senate races. NY-GOV covers the New York governor's race.

Tags do not compound: that is, "education reform" is a completely different tag from "education". A tag like "reform" alone is probably not meaningful.

Consider if one or more of these tags fits your diary: Civil Rights, Community, Congress, Culture, Economy, Education, Elections, Energy, Environment, Health Care, International, Labor, Law, Media, Meta, National Security, Science, Transportation, or White House. If your diary is specific to a state, consider adding the state (California, Texas, etc). Keep in mind, though, that there are many wonderful and important diaries that don't fit in any of these tags. Don't worry if yours doesn't.

You can add a private note to this diary when hotlisting it:
Are you sure you want to remove this diary from your hotlist?
Are you sure you want to remove your recommendation? You can only recommend a diary once, so you will not be able to re-recommend it afterwards.
Rescue this diary, and add a note:
Are you sure you want to remove this diary from Rescue?
Choose where to republish this diary. The diary will be added to the queue for that group. Publish it from the queue to make it appear.

You must be a member of a group to use this feature.

Add a quick update to your diary without changing the diary itself:
Are you sure you want to remove this diary?
(The diary will be removed from the site and returned to your drafts for further editing.)
(The diary will be removed.)
Are you sure you want to save these changes to the published diary?

Comment Preferences

    •  Thank you for this- (3+ / 0-)

      Tipped and recced for showing how and why statistics should be used, and also for phrasing the hypothesis in a proper way!

      Why yes, I am Catholic.

      by ems97007 on Sat Dec 12, 2009 at 04:00:13 PM PST

      [ Parent ]

    •  FYI (0+ / 0-)

      It's actually quite explicable: I posted it because I don't know anything about it and 1) I was interested in what those who do know about it thought, 2) a veteran member brought up in a slightly critical way that I've never posted on a single diary or comment on gun control and add that he rarely sees anything about it on the FP by anyone else. So, don't get me wrong, I'm glad you posted this diary, it's the kind of discussion I hoped to see. But to be clear, it's not that it's inexplicable, it's that you didn't know the reasons and as far as I know you didn't ask me. My apologies if you did and I missed it.

      I think the AGR analyst would be happy to look at your analysis though and link back to you. You can contact him from his homepage.

    •  I have to wonder about the accuracy of the data.. (0+ / 0-)

      as everything I have seen indicates growing firearms ownership rates, not falling.  I tried to look at the info on the GSS web site, but holy bedrock, that site sucks.

  •  I just read (1+ / 0-)
    Recommended by:
    neroden

    That suicide rates are 5 times higher in houses that contain guns than in houses that do not contain guns.

    It was not a scientific article however, so I do not know if that was a hard fact.

  •  Thank you for the stats (4+ / 0-)
    Recommended by:
    bustacap, Fabian, chocolate cliffs, Ojibwa

    I like seeing hard numbers.

    -6.88 eco -6.15 soc 12-12-09 Hist Channel 8-10 PM Howard Zinn -The People Speak

    by CuriousBoston on Sat Dec 12, 2009 at 01:17:47 PM PST

  •  and I'm guessing 120% agree with him. n/t (3+ / 0-)
    Recommended by:
    allep10, Angela Quattrano, Ojibwa

    Everyone dies but very few people really live.

    by thestupiditburns on Sat Dec 12, 2009 at 01:18:35 PM PST

  •  I didn't expect that. (3+ / 0-)
    Recommended by:
    wu ming, bustacap, Ojibwa

    I'd be very interested in murder-suicide numbers as well.  I'd expect they track more closely with unemployment.

    It looks like suicidal thoughts + guns => suicide  more often than suicidal thoughts + no guns.  

    Show me the POLICY!

    by Fabian on Sat Dec 12, 2009 at 01:22:54 PM PST

  •  I also took a look at that link (7+ / 0-)

    I'm always eager to find solid, evidence-based analysis of risks and benefits associated with firearm ownership, and I eagerly followed that link.

    I was extremely disappointed. I saw no science or objective analysis there.

    DarkSyde's links are usually a lot better than that.

    •  Indeed. I totally agree and was equally (1+ / 0-)
      Recommended by:
      GreenHills

      disappointed when I followed the link.

      I'd also like to see statistical support that the longer it takes one to find the means of comitting suicide has any correlation with whether an individual goes ahead and actually commits suicide.

      Most studies I've looked at suggest if someone really wants to commit suicide, they'll find a way.

  •  Not quite sure what you're trying to say... (2+ / 0-)
    Recommended by:
    greengemini, allep10

    Right up front: I'm not a statistician.

    I think what you're saying comports with my intuitive response to the question, though: the close availability of a gun has more of an impact on the suicide rate than does unemployment.  And that's not surprising.

    More often than not, committing suicide is a very emotional response to an acute condition brought on by cascading factors over a brief period of time (ie., foreclosure papers were served, spouse walked out because of that, started drinking, a sad country western song came on the ipod shuffle, and wham.  

    If there's a firearm close at hand, the decision can be too easy, and too final once the trigger is pulled.

    On the other hand, unemployment is more of chronic condition that, absent other cascading events, is unlikely to lead to either self-poisoning (pills) or a quick solution (a gun).

    There is a fine line between bold and Dennis Kucinich.

    by Richard Cranium on Sat Dec 12, 2009 at 01:41:06 PM PST

  •  Looking at the chart, there may be (0+ / 0-)

    a significant correlation between non-firearm suicides and unemployment.

    Did you, and if not could you run that calculation?

    The most important way to protect the environment is not to have more than one child.

    by nextstep on Sat Dec 12, 2009 at 01:41:20 PM PST

    •  Yes, and no (1+ / 0-)
      Recommended by:
      neroden

      Yes, I ran the correlation.

      No, it's not significant. The correlation coefficient r between nongun suicides and uemployment is 0.252, nowhere near significance.

      If you don't stand for something, you'll stand for anything.

      by Keith Pickering on Sat Dec 12, 2009 at 01:44:31 PM PST

      [ Parent ]

      •  What was t and Prob(t)? (0+ / 0-)

        The most important way to protect the environment is not to have more than one child.

        by nextstep on Sat Dec 12, 2009 at 02:15:31 PM PST

        [ Parent ]

        •  A bit of explaination (1+ / 0-)
          Recommended by:
          bustacap

          Each variable has it's own error associated with it. Basically, we jiggle the variable a little and see how much the model changes when we jiggle. If the model doesn't change much, that means we really don't know much about the variable, in other words, the error of that variable is high. But if the model changes a lot when we jiggle, that ties the variable down to a much narrower range, otherwise the model won't fit.

          So if you look at the "estimate", that's what we're saying is the best guess for the variable, and the "error" tells us how confident we are in the variable. If the error is small compared to the estimate, that's good! It tells us that there's "really something there". But if the error is large compared to the variable, that's bad, because it implies that the variable isn't telling us anything.

          The t statistic is simply the ratio of the estimate to its error. The higher the t statistic, the more confident we are that the variable is "real", i.e., that it's contributing to real effects in the model. In this case, the t-statistic for gun ownership is almost 15, which is huge. The t-statistic for unemployment is –1.4, which is quite small. (You look at absolute value with the t-statistic, so –15 would have been huge too. Anything between -2 and 2 is pretty dicey, in my experience.)

          The prob(t) tells you the probability that the variable might actually be zero. If the variable is zero, it's not contributing anything to the model: the variable just isn't related to the dependent varaible (suicides, in this case). If the prob(t) isn't less than 5% or 10%, it's not statistically significant and can be safely eliminated from the model.

          If you don't stand for something, you'll stand for anything.

          by Keith Pickering on Sat Dec 12, 2009 at 02:26:23 PM PST

          [ Parent ]

          •  I should have been more precise (0+ / 0-)

            in my question.  What was t and Prob(t) for non-firearm suicides?

            The most important way to protect the environment is not to have more than one child.

            by nextstep on Sat Dec 12, 2009 at 02:29:55 PM PST

            [ Parent ]

            •  Sorry ... (1+ / 0-)
              Recommended by:
              AaronInSanDiego

              Since we didn't include non-firearm suicides in the model, we didn't calculate t or prob(t). In looking at the correlation matrix, nongun suicides are not correlated with anything at a statistically significant level, so it's not really useful to build a model with that included. At least not with firearms or unemployment.

              If you don't stand for something, you'll stand for anything.

              by Keith Pickering on Sat Dec 12, 2009 at 02:36:56 PM PST

              [ Parent ]

  •  Whichever side is right (3+ / 0-)
    Recommended by:
    Kimberley, Big Tex, GreenHills

    it shouldn't matter in the debate on whether the people have a RIGHT to own a gun.  Just because the right can result in untoward consequences (exclusionary rule and the guilty going free, anyone?) does not mean that the right doesn't exist.

    •  It does have an effect (0+ / 0-)

      ... on policy decision, though. The right to own a gun is not, and should not be, absolute. This is one reason why.

      If you don't stand for something, you'll stand for anything.

      by Keith Pickering on Sat Dec 12, 2009 at 02:08:45 PM PST

      [ Parent ]

    •  Agreed (0+ / 0-)

      The First Amendment's currently being misused by overly powerful interests to slit the country's throat and I've got a knuckle sandwich for the first person who says we ought to get rid of that.

      •  Oh dear. (0+ / 0-)

        The problem is, how do you reign in those interests without infringing on what they call their freedom? You can't talk louder then them, because they own the machinery for loud talking. So you either dial them down a notch, or shoot them, or surrender to them. Those are the basic choices, as I see them.

        I come down on the side of less power and therefore less freedom for them.

        A soft answer turneth away wrath: but grievous words stir up anger (Proverbs 15:1)

        by Boreal Ecologist on Sat Dec 12, 2009 at 03:25:16 PM PST

        [ Parent ]

  •  One number that's not static is age of gun owners (3+ / 0-)
    Recommended by:
    Big Tex, rodentrancher, GreenHills

    What I'm getting at is the suicide rate with firearms is much higher in the late middle age bracket than amongst other ages.

    Perhaps it would be necesary to adjust for the aging population.

    Not the topic of this diary but I'd suggest anyone bringing a gun into the house take a real serious look at the likelihood of suicide of any household member as well as anger.

    an accounting of daily gun deaths

    "Don't fall or we both go" Derek Hersey

    by ban nock on Sat Dec 12, 2009 at 02:11:04 PM PST

  •  Note that the age distribution greatly changed (4+ / 0-)

    over those years.  Those over 65 years of age have a 50% higher suicide rate than the general population.

    So, if all other things stayed the same over those years, we would expect the suicide rate to go up because the elderly comprise an increasing higher fraction of the population.

    Therefore this factor needs to adjust the suicide rates before looking at the effects of firearm ownership or unemployment.

    The most important way to protect the environment is not to have more than one child.

    by nextstep on Sat Dec 12, 2009 at 02:14:06 PM PST

    •  Also one should look at (1+ / 0-)
      Recommended by:
      Fabian

      rates of treatment of mental illness and increasing types of medication available. There are a lot of factors that I think have more to do with changing suicide rates than unemployment. On the other hand, unemployment can affect rates of mental illness by changing the number of people covered by medical insurance.

      Economic Left/Right: -4.00 Social Libertarian/Authoritarian: -6.82

      Your argument is not Scottish.

      by AaronInSanDiego on Sat Dec 12, 2009 at 06:35:43 PM PST

      [ Parent ]

  •  Bear with me because I'm not a statistician, (0+ / 0-)

    but it seems problematic to say that gun ownership tracks closely with suicide rates while unemployment doesn't, when you're taking gun ownership rates and multiplying them by 0.2.  I understand that this is being done for purposes of comparison, but it seems that this would understate fluctuations in the data and make the graph line flatter than it should be.  I guess I'm wondering why you would do this with the gun ownership data but not the unemployment data, or why it wouldn't be better to just overlay the unadjusted graph lines to show how closely they track or don't track with onee another.

    As for the stuff about null hypothesis, that was completely over my head and I'm not touching it with a ten-foot pole.  Or poll.

    -7.12, -7.54 / "Health care reform will never take place until Rahm Emanuel is strangled with the entrails of Frank Luntz." - Diderot

    by Big Tex on Sat Dec 12, 2009 at 02:21:39 PM PST

    •  uh, no. (1+ / 0-)
      Recommended by:
      AaronInSanDiego

      I multiplied by 0.2 only to fit the lines on the graph nicely. It has no effect on the statistics.

      As a matter of fact, if you have two datasets you're trying to correlate and multiply either one, or both, by any constant, that has absolutely no effect on the correlation coefficient. None. (Unless you multiply by zero, of course).

      If you don't stand for something, you'll stand for anything.

      by Keith Pickering on Sat Dec 12, 2009 at 02:41:40 PM PST

      [ Parent ]

      •  Well, like I said, (0+ / 0-)

        I'm not a statistician and I'm not entirely sure what the correlation coefficient even is, or its relevance in this case.  So be patient with me.  What I'm saying is that multiplying by 0.2 as you've done here, has a definite effect on the way that the line appears on the graph.  Here's a comparison between the gun ownership numbers before and after multiplying by 0.2:

        Graph

        As you can see, the line representing the adjusted numbers is a lot flatter than the one representing the unadjusted numbers.  As a non-statistician, maybe there's something lurking in the gap between my level of understanding and yours that explains why it makes sense, in a visual representation like this, to multiply the gun ownership number by a constant but not the unemployment numbers, or anything else.  But it just seems that it would distort the visualization of these numbers in a way that's misleading.  Why multiply one set of numbers by a constant instead of just overlaying the graph lines?  (I'm asking partially out of ignorance here, not just to challenge your representation of these numbers.)

        -7.12, -7.54 / "Health care reform will never take place until Rahm Emanuel is strangled with the entrails of Frank Luntz." - Diderot

        by Big Tex on Sat Dec 12, 2009 at 04:10:48 PM PST

        [ Parent ]

        •  Appears is the operative word. (2+ / 0-)
          Recommended by:
          AaronInSanDiego, bustacap

          The only reason I multiplied by 0.2 was to get the line to fit nicely on the graph. I did NOT fiddle with the underlying numbers when doing the statisical analysis -- although it wouldn't have made any difference if I had, since correlation is looking at the variances, which do not change relative to the mean after any multiplication process.

          Here's an example. Consider the following two datasets:
          Set A
          1 1.5 2 3 8
          Set B
          10 15 20 30 80

          Set B is just set A multiplied by 10. These two datasets have a correlation coefficient of 1.0, perfect correlation. Set A also correlates with itself by 1.0, perfect correlation. In fact, if you multiply Set A by any number except zero, you will generate a dataset that correlates with Set A perfectly.

          If you don't stand for something, you'll stand for anything.

          by Keith Pickering on Sat Dec 12, 2009 at 04:17:51 PM PST

          [ Parent ]

          •  Okay, I think I understand a little better now. (1+ / 0-)
            Recommended by:
            AaronInSanDiego

            Your example made things a little clearer for me.  Thanks for the impromptu math tutorial.  :-)

            -7.12, -7.54 / "Health care reform will never take place until Rahm Emanuel is strangled with the entrails of Frank Luntz." - Diderot

            by Big Tex on Sat Dec 12, 2009 at 04:49:31 PM PST

            [ Parent ]

  •  As a long time (0+ / 0-)

    gun owner and hunter your theory doesn't wash.Anyone who has seen the damage a 12ga shotgun does to flesh,animal or human knows that there are better and less messy ways to commit suicide.Most gun owners would seek a easier more effective way to kill themselves.Also shooting yourself is not always effective in the mission to complete the job,you could end up living through your attempt and face the burden you would leave to your family to clean up the mess you made.Killing yourself with a gun would be made by a novice not a experienced  gun owner.

    •  Then perhaps you can explain ... (4+ / 0-)

      ... why about half of all suicides are by firearm? It strikes me that nothing is easier, and nothing is more effective, than a gun, if you want to kill yourself.

      If you don't stand for something, you'll stand for anything.

      by Keith Pickering on Sat Dec 12, 2009 at 02:43:15 PM PST

      [ Parent ]

    •  Then how do you explain the data? (1+ / 0-)
      Recommended by:
      bustacap

      I suppose higher levels of gun ownership could make people more inclined to poison or hang themselves. Is that the mechanism you are proposing to explain the data?

      A soft answer turneth away wrath: but grievous words stir up anger (Proverbs 15:1)

      by Boreal Ecologist on Sat Dec 12, 2009 at 03:28:11 PM PST

      [ Parent ]

    •  Suicide rates are higher in rural areas, (1+ / 0-)
      Recommended by:
      bustacap

      if I correctly remember a New York Times article, and also stunningly high for men in their 40s.  Rural suicide rates are thought to be related to the lower availability of mental health services, and higher expectation for independent perseverence.  Since rural populations are decreasing, one might expect both gun ownership to decrease AND suicide rates.  So I too question whether the ownership is directly related to suicide or whether both correlating variables are driven by other factors (rural cultural factors).

    •  Wow, there's a metric buttload of (0+ / 0-)

      ... conclusion jumping and projection in that post, lefty.  There's this logical fallacy called argument from omniscience that I believe fits your assertion.  Trouble is, your claim doesn't fit the data at all.

      ... and we have seen the black suns | pouring forth the night. -- Clark Ashton Smith

      by bustacap on Sat Dec 12, 2009 at 06:25:56 PM PST

      [ Parent ]

  •  Another thought: (1+ / 0-)
    Recommended by:
    JayH

    This is directed more at the Brady Institute's assertion that increased gun ownership causes more suicides - without being able to review the methodology by which they came to their conclusion, it nevertheless seems a pretty reckless assertion on their part.  Since it's generally understood that using a gun is one of the most efficient and effective ways to commit suicide, it seems that if a person were preoccupied with suicide, they might be more likely to go out and purchase a gun for the purpose of killing themself.  Is it really fair to assume that the presence of a gun leads to suicide, when it might be the other way around - suicidal thoughts leading to the purchase of a gun?  And if a person is sufficiently inclined to kill themself that they would use a gun to do so, wouldn't they just find another method of suicide if they weren't able to obtain a gun?

    -7.12, -7.54 / "Health care reform will never take place until Rahm Emanuel is strangled with the entrails of Frank Luntz." - Diderot

    by Big Tex on Sat Dec 12, 2009 at 02:39:51 PM PST

    •  Impossible (0+ / 0-)

      When one household out of three owns a gun, and one person out of 10,000 commits suicide in any given year, the number of purchased-for-suicide guns has an utterly trivial impact on the overall ownership rate.

      On the other hand, if you're depressed and a gun is available, you're a lot more likely to kill yourself than if you're depressed and a gun is not available.

      If you don't stand for something, you'll stand for anything.

      by Keith Pickering on Sat Dec 12, 2009 at 02:47:32 PM PST

      [ Parent ]

      •  ... (0+ / 0-)

        When one household out of three owns a gun, and one person out of 10,000 commits suicide in any given year, the number of purchased-for-suicide guns has an utterly trivial impact on the overall ownership rate.

        But since the numbers for gun ownership are so high, and the numbers for suicide are so low, and given the existence of other potential factors that might affect both the suicide rate and the gun ownership rate, isn't it unsound to attempt to draw a correlation between the two as the Brady people have done?  Especially when a significant percentage of the people who commit suicide do so without firearms?

        On the other hand, if you're depressed and a gun is available, you're a lot more likely to kill yourself than if you're depressed and a gun is not available.

        You're more likely to succeed in killing yourself, but that doesn't mean that you're more likely to attempt or want to do so.  The Brady people seem to be suggesting that gun ownership leads to suicide, which is a reach.

        -7.12, -7.54 / "Health care reform will never take place until Rahm Emanuel is strangled with the entrails of Frank Luntz." - Diderot

        by Big Tex on Sat Dec 12, 2009 at 04:31:05 PM PST

        [ Parent ]

        •  So if you want to fly (0+ / 0-)

          and have no wings, you're less likely to succeed than if you want to fly and do have wings.

          I think this is a semantic quibble rather than a statistical one. If you don't want to say that guns cause suicide, it certainly is fair to say that guns allow suicides that otherwise would not have succeeded.

          If you don't stand for something, you'll stand for anything.

          by Keith Pickering on Sat Dec 12, 2009 at 04:34:38 PM PST

          [ Parent ]

          •  Of course they do, (1+ / 0-)
            Recommended by:
            theatre goon

            but that's like saying that you're more likely to die driving from point A to point B than, say, riding a bike.  If there are more people driving then there will probably be more people dying in car crashes, but the actual cause of the fatality isn't the car itself or the act of driving - it might be drunk driving, or lack of sleep, or a patch of ice on the road, or any number of other things.  Likewise, if someone uses a gun to kill themself, it's not the gun causing them to do so - it's the fact that they're suffering from untreated depression, or going through some sort of personal crisis, or whatever.

            The Brady people are using whatever correlation exists between gun ownership and suicide rates as a talking point in a campaign to severly restrict gun ownership rights.  What they're doing is the equivalent of trying to minimize car crash fatalities by severely restricting the right of people to own cars.

            -7.12, -7.54 / "Health care reform will never take place until Rahm Emanuel is strangled with the entrails of Frank Luntz." - Diderot

            by Big Tex on Sat Dec 12, 2009 at 05:12:21 PM PST

            [ Parent ]

  •  Owning gun=more likely to shoot self? (1+ / 0-)
    Recommended by:
    bustacap

    Who knew?

    This should fall under the category of well, Duh.

  •  It is obvious why gun ownership was left out. (2+ / 0-)
    Recommended by:
    murrayewv, Keith Pickering

    It would have made the firearm-suicide data much harder to see by covering up and obscuring so many of the datapoints.

    illegal, n. A term used by descendents of European immigrants to refer to descendants of Indigenous Americans

    by ricardomath on Sat Dec 12, 2009 at 02:45:06 PM PST

  •  The correlation is probably valid, but (2+ / 0-)
    Recommended by:
    GreenHills, JayH

    I'm not sure how much it means.

    It may mean no more than that, all other things being equal, a gun-owner who is intent on committing suicide is likely to do it with a gun, while a non-gun-owner is far more likely not to do it with the gun he doesn't own.

    So if gun ownership goes up, use of guns among suicidal people is going to go up; if gun ownership goes down, suicidal people without guns will likely find some other means of committing suicide.

    That doesn't lead one to the conclusion that having a gun leads people to commit suicide.

    •  In that case ... (0+ / 0-)

      How about non-gun suicides? Which remain flat regardless of the gun ownership rates?

      That implies that non-gun owners are committing suicides at a rate which has nothing do do with gun ownership.

      If some endogenous factor were causing suicides, then as gun ownership declined, non-gun suicides should be rising as non-gun owners found other ways to commit suicide. That's not happening. The only conclusion is that gun ownership leads to suicide. Not causes, necessarily, but leads to.

      If you don't stand for something, you'll stand for anything.

      by Keith Pickering on Sat Dec 12, 2009 at 03:28:14 PM PST

      [ Parent ]

  •  I own a gun and I'm unemployed (0+ / 0-)

    I'm thinking of hanging myself, poisoning myself or jumping off a cliff

    Happy just to be alive

    by exlrrp on Sat Dec 12, 2009 at 03:18:24 PM PST

  •  Data? (0+ / 0-)

    A fascinating analysis!

    Could you post the raw data you assembled for the analysis?  Right here as text in a comment would be great.  I think the data would be instructive for HS/college classes to study.

    Thanks for this diary!

    •  Here it is (0+ / 0-)

      Data sources listed in the diary.

      Year Unemployment Firearm suicide non-firearm suicide Total suicide Firearm % Gun ownership
      1982 9.7 9.12 6.42 15.54 58.69 9.78
      1984 7.5 9.22 6.53 15.75 58.54 9.7
      1985 7.2 9.24 6.41 15.65 59.04 9.62
      1987 6.2 9.44 6.59 16.03 58.89 9.72
      1988 5.5 9.39 6.3 15.69 59.85 8.68
      1989 5.3 9.32 6.17 15.49 60.17 9.78
      1990 5.6 9.59 6.09 15.68 61.16 9.16
      1991 6.8 9.29 6.15 15.44 60.17 8.74
      1993 6.9 9.24 5.92 15.16 60.95 9.1
      1994 6.1 9.04 5.96 15.00 60.27 8.78
      1996 5.4 8.54 5.97 14.51 58.86 8.68
      1998 4.5 7.53 5.99 13.52 55.70 7.34
      2000 4.0 7.45 5.68 13.13 56.74 6.86
      2002 5.8 7.49 6.32 13.81 54.24 7.28
      2004 5.5 7.19 6.67 13.86 51.88 7.46
      2006 4.6 7.08 6.84 13.92 50.86 6.9

      If you don't stand for something, you'll stand for anything.

      by Keith Pickering on Sat Dec 12, 2009 at 03:49:59 PM PST

      [ Parent ]

      •  You will notice (0+ / 0-)

        that one column above, "Firearm %", is not used in my analysis. I didn't find much interesting in it, but it was in my spreadsheet so I'm including it here too. You can compute it yourself with the underlying firearm/non-firearm data.

        If you don't stand for something, you'll stand for anything.

        by Keith Pickering on Sat Dec 12, 2009 at 05:17:00 PM PST

        [ Parent ]

      •  CAUTION (0+ / 0-)

        I just realized that the portion of the spreadsheet I took the above post from was the part I used to generate the graph. So the gun ownership rates have been multiplied by 0.2 as the graph shows. To generate the real underlying GSS data, you need to multiply the final column by 5.

        If you don't stand for something, you'll stand for anything.

        by Keith Pickering on Sat Dec 12, 2009 at 05:19:20 PM PST

        [ Parent ]

  •  A gun owner's thought on gun suicides (2+ / 0-)
    Recommended by:
    bustacap, Fabian

    Practically all gun suicides happen when depressed, drunk, or drugged people get their hands on a gun. (No, I don't have statistics, but come on, how many happy, healthy, straight, sober people kill themselves?)

    Regardless of what you think about the gun control debates, you have a moral responsibility as a gun owner.  If someone in your household is a problem drinker, doing drugs, suffering from mental illness, or otherwise badly troubled, you need to get the guns out of the house, now! Guns and troubled people are a very bad combination.

    On the other hand, given that no one in my home is depressed, drinks, or does drugs, you can hardly expect me to give up my bird gun or deer rifle because other people have used their guns to kill themselves.

  •  Excellent argument (0+ / 0-)

    It completely supports loosening of gun laws so that more people have firearm possession opportunities. Only a right-wing religious fanatic would try to deny the individual right to off themselves in an efficient method with a high degree of predictable success.

    Guns are Good!!!

  •  One quibble... (2+ / 0-)
    Recommended by:
    Big Tex, JayH

    It turns out that the Brady Institute is right...

    Not so much.  As per the diarist, what the Brady Institute says is that "increased gun ownership causes more suicides."  Emphasis added.

    This does not show that gun ownership causes suicide -- any more than having a driver's license causes drunk driving.  It only shows that those who are already suicidal and own a gun are more likely to use that gun to commit suicide, not that one who owns a gun is more likely to commit suicide.

    If you dig zombies, and get the chance, go here and vote for Video#10. Thanks.

    by theatre goon on Sat Dec 12, 2009 at 04:06:23 PM PST

    •  Formally correct, but... (1+ / 0-)
      Recommended by:
      AaronInSanDiego

      ... when 1/3 of households own a gun, and 50% to 60% of suicides are by firearm, then yes, owning a gun does make suicide more likely.

      If you don't stand for something, you'll stand for anything.

      by Keith Pickering on Sat Dec 12, 2009 at 04:09:38 PM PST

      [ Parent ]

      •  Still, the same basic quibble. (3+ / 0-)
        Recommended by:
        AaronInSanDiego, Big Tex, JayH

        You seem to be asserting a causal relationship, while the analysis only shows a correlational relationship.

        In other words, if a gun is available, it is more likely to be used to commit suicide by a suicidal person -- the gun does not cause a person to become suicidal.

        I still believe the driver's license/drunk driving example is appropriate here -- the vast majority of people who drive drunk also have a driver's license, but they license did not cause them to drive drunk.

        If you dig zombies, and get the chance, go here and vote for Video#10. Thanks.

        by theatre goon on Sat Dec 12, 2009 at 04:14:57 PM PST

        [ Parent ]

        •  A gun (2+ / 0-)
          Recommended by:
          AaronInSanDiego, bustacap

          A gun doesn't cause suicidal thoughts. BUT, it does allow suicidal thoughts to be exercised very easily.

          You are correct that correlation is not causation. But knowing that does not mean that we can just turn off our brains and pretend that there cannot be causation. So let's think it through. When we have linkage, there are three possiblities: A causes B, B causes A, or something else causes both A and B.

          Gun ownership is strongly correlated with suicide. In an earlier comment, Big Tex suggested that suicidal people are more likely to go out and buy guns. And while that may be theoretically true, as I pointed out in my response, when 1 in 3 homes owns a gun but only 1 in 10,000 people commits suicide in any given year, it is clear that suicidal people in aggregate can have no measurable effect on the rates of gun ownership. There are just not enough of them to make an impact. In fact, the gun ownership data is only captured at the 0.1% level, which is an order of magnitude too low to even theoretically measure the effect.

          So no, we can say unequivocally that suicides cannot possibly have any effect on the rates of gun ownership. Or if they do, it's too small to measure and cannot be (and has not been) captured in the statistics.

          That leaves one more possibility: that there is "something else" that causes both gun ownership and suicidal thoughts. On its face, this seems quite unlikely. But you're free to argue it if you wish. I just can't for the life of me figure out what "something else" might be.

          If you don't stand for something, you'll stand for anything.

          by Keith Pickering on Sat Dec 12, 2009 at 04:29:03 PM PST

          [ Parent ]

          •  Seems pretty clear. (2+ / 0-)
            Recommended by:
            Big Tex, Fabian

            Suicides tend to be committed by whatever means are at hand -- as you mention when referring to Big Tex, people don't tend to go out and buy something to commit suicide with.

            If there is an increase in gun ownership, then that means there are more at hand, and therefore guns are used more often in the commission of suicide.  The two are correlated, not causal.

            None of this would lead to a reasonable assumption that there is "something else" that causes both gun ownership and suicidal thoughts, only than when a gun happens to be owned, it is the more likely vehicle used to commit suicide.

            If you dig zombies, and get the chance, go here and vote for Video#10. Thanks.

            by theatre goon on Sat Dec 12, 2009 at 04:36:29 PM PST

            [ Parent ]

            •  Not just that. (2+ / 0-)
              Recommended by:
              AaronInSanDiego, Fabian

              Guns are very lethal, so when you choose a gun for suicide, you're much more likely to succeed. So yes, guns do cause suicide, because if you're using a gun, you will succeed, while if you're using pills you're usually going to fail.

              If you don't stand for something, you'll stand for anything.

              by Keith Pickering on Sat Dec 12, 2009 at 04:42:37 PM PST

              [ Parent ]

              •  This is the key (0+ / 0-)

                for an individual suicide involving a gun, there is a definite causal relationship. That combined with the statistical correlation presents a pretty clear case in my view. Unless there's another explanation, which I haven't seen yet.

                Economic Left/Right: -4.00 Social Libertarian/Authoritarian: -6.82

                Your argument is not Scottish.

                by AaronInSanDiego on Sat Dec 12, 2009 at 06:42:25 PM PST

                [ Parent ]

              •  And yet... (0+ / 0-)

                ...the gun is not causing the suicide -- the person committing the suicide is.

                That's what it really comes down to, and that being the case, guns do not cause suicide -- suicidal people do.  They very well may facilitate it, but they do not cause it, any more than a car causes drunk driving -- even though you have to have one to do it.

                If you dig zombies, and get the chance, go here and vote for Video#10. Thanks.

                by theatre goon on Sat Dec 12, 2009 at 07:47:03 PM PST

                [ Parent ]

                •  The gun is the proximate cause of the death (1+ / 0-)
                  Recommended by:
                  Fabian

                  to say otherwise doesn't make sense.

                  Economic Left/Right: -4.00 Social Libertarian/Authoritarian: -6.82

                  Your argument is not Scottish.

                  by AaronInSanDiego on Sat Dec 12, 2009 at 11:48:46 PM PST

                  [ Parent ]

                •  With regard to drunk driving (0+ / 0-)

                  you have two ingredients. Alcohol is the cause of a person being drunk, a car is the tool with which a person drives. Both ingredients are necessary and sufficient to cause drunk driving. With suicide by firearm, the firearm is the only element necessary. You might say that suicidal intent is also necessary, but that is not an easily quantifiable material object, as alcohol, cars, and guns are.

                  Economic Left/Right: -4.00 Social Libertarian/Authoritarian: -6.82

                  Your argument is not Scottish.

                  by AaronInSanDiego on Sat Dec 12, 2009 at 11:54:01 PM PST

                  [ Parent ]

                  •  That makes no sense... (0+ / 0-)

                    You might say that suicidal intent is also necessary...

                    Of course this is necessary, and clearly the only required ingredient -- otherwise, you are claiming that an otherwise non-suicidal person will, only because of the presence of a gun, suddenly become suicidal, which is more ludicrous than my purposefully ludicrous example of a car causing drunk driving.

                    Or, I suppose you could say that if someone hangs themselves, then it was the rope that caused them to commit suicide, not, for instance, untreated clinical depression.

                    If all suicides were committed with firearms, there would at least be some evidence to what you are claiming, or if every gun owner eventually committed suicide.  Otherwise, the firearm is still just the means the suicide has chosen to carry out their intent.

                    If you dig zombies, and get the chance, go here and vote for Video#10. Thanks.

                    by theatre goon on Sun Dec 13, 2009 at 05:06:55 AM PST

                    [ Parent ]

                    •  Yes, of course the intent is necessary (0+ / 0-)

                      but there are many people who have various levels of suicidal ideation who don't successfully perform the act. The diarist's correlation was specifically with suicide by firearm. There was a poor correlation between non-firearm suicide rates and gun ownership rates. I assume the overall suicide rate correlated somewhat because firearms are a common method for suicide in this country. It would be interesting to look at other methods as well, such as correlating rate of suicide by overdose of prescription drugs with rates of prescriptions for those drugs, or the number and height of bridges with bridge suicides.

                      Economic Left/Right: -4.00 Social Libertarian/Authoritarian: -6.82

                      Your argument is not Scottish.

                      by AaronInSanDiego on Sun Dec 13, 2009 at 08:38:18 AM PST

                      [ Parent ]

                      •  And... (0+ / 0-)

                        ...the diarist went on to say that firearms cause suicides -- a causal relationship, when it is a correlational relationship at best, as I've already pointed out.

                        So, what we arrive at is that guns do not, in fact, cause suicide, but that when suicide is intended, gun owners tend to use a firearm rather than some other method -- quite a distance from what the Brady group was claiming, that it is the gun that causes the suicide, rather than that the gun is the means of that suicide, when they are already owned.

                        Not news, and the Brady claims are, as they are so often shown to be, incorrect.

                        If you dig zombies, and get the chance, go here and vote for Video#10. Thanks.

                        by theatre goon on Sun Dec 13, 2009 at 09:13:04 AM PST

                        [ Parent ]

                        •  The fact that guns are directly involved in (0+ / 0-)

                          the suicides we're talking about, and the gunshot wound is the proximate cause of death, establishes the causal relationship. If there were a correlation between gun ownership and non-gun suicides, your argument would be correct. But the correlation is not all we're dealing with here. We'd have to look at other methods of suicide to determine if guns are more likely to cause suicides than other implements.

                          Economic Left/Right: -4.00 Social Libertarian/Authoritarian: -6.82

                          Your argument is not Scottish.

                          by AaronInSanDiego on Sun Dec 13, 2009 at 09:28:07 AM PST

                          [ Parent ]

                          •  Still wrong. (0+ / 0-)

                            A causal relationship would be one in which the gun is the cause of the suicide -- the thing that causes one to commit suicide, not just the tool used.

                            A better example is something like untreated clinical depression, which actually does, at times, cause suicide.  A causal relationship.

                            A correlational relationship is one in which, when suicide is committed, if a firearm is in the home, that is the tool used.  The firearm itself does not cause the suicide, it is merely the tool used.  Unless the gun were somehow to influence the decision to commit suicide -- an impossibility, as it is an inanimate object -- there is no causal relationship.

                            If you dig zombies, and get the chance, go here and vote for Video#10. Thanks.

                            by theatre goon on Sun Dec 13, 2009 at 01:34:33 PM PST

                            [ Parent ]

                          •  I think we're going in circles on this (0+ / 0-)

                            and I doubt we'll convince each other. I will say that I think it depends on whether you are talking specifically about suicide by firearm or suicide generally. I think that the availability of a gun is a causal factor in gun suicide, because without the gun there would not be suicide by gun. That's not saying there wouldn't be suicide by other means. Obviously that's still a possibility, but the question in my mind is, how does the presence of a gun affect the likelihood of suicide? That's something that would require more information to reach a conclusion about. But the strong correlation between gun ownership and gun suicide rates suggests that it may make successful suicide more likely.

                            Economic Left/Right: -4.00 Social Libertarian/Authoritarian: -6.82

                            Your argument is not Scottish.

                            by AaronInSanDiego on Sun Dec 13, 2009 at 07:58:02 PM PST

                            [ Parent ]

                          •  We actually agree... (0+ / 0-)

                            ...on how you've expressed it this time, more or less.

                            The major difference we have is in what is a causal and what is a correlational factor.  A causal factor is the underlying cause of the suicide -- i.e., why someone commits suicide.

                            A correlational factor is one such as this one, a suicidal person who happens to own a firearm may be more likely to use that as a means to commit suicide than another means.

                            The gun itself is not the underlying cause of the suicide any more than gravity is the reason someone decides to jump off a building.

                            I think that the availability of a gun is a causal factor in gun suicide, because without the gun there would not be suicide by gun.

                            This is a self-fulfilling prophecy of sorts -- of course if one decides to shoot themselves, they use a gun.  It does not, however, tell us anything about the why, only the how.

                            If you dig zombies, and get the chance, go here and vote for Video#10. Thanks.

                            by theatre goon on Mon Dec 14, 2009 at 04:36:46 AM PST

                            [ Parent ]

  •  Quite a bit of evidence..... (3+ / 0-)

    shows a higher suicide rate in states where there are more guns.

    In the first nationally representative study to examine the relationship between survey measures of household firearm ownership and state level rates of suicide in the U.S., researchers at the Harvard School of Public Health (HSPH) found that suicide rates among children, women and men of all ages are higher in states where more households have guns. The study appears in the April 2007 issue of The Journal of Trauma.

    "We found that where there are more guns, there are more suicides," said Matthew Miller, Assistant Professor of Health Policy and Management at HSPH and lead author of the study.

    Suicide ranks as one of the 15 leading causes of death in the U.S.; among persons less than 30 years old, it is one of the top three causes of death. In 2004, more than half of the 32,439 Americans who committed suicide used a firearm.

    link

    You shall know the truth, and the truth shall make you mad. Aldous Huxley

    by murrayewv on Sat Dec 12, 2009 at 04:55:22 PM PST

    •  More evidence..... (3+ / 0-)

      Ready availability of firearms is associated with an increased risk of suicide in the home. Owners of firearms should weigh their reasons for keeping a gun in the home against the possibility that it might someday be used in a suicide.

      If only unemployment affected suicide, more young black US men would commit suicide.  But young white men commit suicide at greater rates.

      You shall know the truth, and the truth shall make you mad. Aldous Huxley

      by murrayewv on Sat Dec 12, 2009 at 05:01:22 PM PST

      [ Parent ]

      •  More information from New England Journal.... (3+ / 0-)

        of Medicine.  This editorial was written in response to the Supreme Court overturning the handgun ban in DC.

        Why might the availability of firearms increase the risk of suicide in the United States? First, many suicidal acts — one third to four fifths of all suicide attempts, according to studies — are impulsive. Among people who made near-lethal suicide attempts, for example, 24% took less than 5 minutes between the decision to kill themselves and the actual attempt, and 70% took less than 1 hour.2

        Second, many suicidal crises are self-limiting. Such crises are often caused by an immediate stressor, such as the breakup of a romantic relationship, the loss of a job, or a run-in with police. As the acute phase of the crisis passes, so does the urge to attempt suicide. The temporary nature and fleeting sway of many suicidal crises is evident in the fact that more than 90% of people who survive a suicide attempt, including attempts that were expected to be lethal (such as shooting oneself in the head or jumping in front of a train), do not go on to die by suicide. Indeed, recognizing the self-limiting nature of suicidal crises, penal and psychiatric institutions restrict access to lethal means for persons identified as potentially suicidal.

        Third, guns are common in the United States (more than one third of U.S. households contain a firearm) and are lethal. A suicide attempt with a firearm rarely affords a second chance. Attempts involving drugs or cutting, which account for more than 90% of all suicidal acts, prove fatal far less often.

         link

        There are at least a dozen U.S. case–control studies in the peer-reviewed literature, all of which have found that a gun in the home is associated with an increased risk of suicide. The increase in risk is large, typically 2 to 10 times that in homes without guns, depending on the sample population (e.g., adolescents vs. older adults) and on the way in which the firearms were stored. The association between guns in the home and the risk of suicide is due entirely to a large increase in the risk of suicide by firearm that is not counterbalanced by a reduced risk of nonfirearm suicide. Moreover, the increased risk of suicide is not explained by increased psychopathologic characteristics, suicidal ideation, or suicide attempts among members of gun-owning households.

        So the studies listed are flawed since they fail to consider the literature of other studies indicating the relationship between gun ownership and increased suicide.

        You shall know the truth, and the truth shall make you mad. Aldous Huxley

        by murrayewv on Sat Dec 12, 2009 at 05:05:53 PM PST

        [ Parent ]

  •  Excellent (1+ / 0-)
    Recommended by:
    Fabian

    It's nice to see people who really understand this stuff present their skills. I work with people who correlations like this for a living, as well as much more complex analysis, and have dabbled in it myself to a degree. Science works.

    Economic Left/Right: -4.00 Social Libertarian/Authoritarian: -6.82

    Your argument is not Scottish.

    by AaronInSanDiego on Sat Dec 12, 2009 at 06:38:44 PM PST

  •  We need Nate Silver (0+ / 0-)

    Please do an analysis on 538.com!

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site