Tuesday, December 18, 2012

Evidence for substitution among murder weapons: Australia


An article in the Washington Post by Matt Miller today argued we should have gun buybacks to confront the problem of violent crime. He cites the example of Australia: in 1996, after a massacre of 35 people, the country banned a broad category of guns and the government repurchased them from citizens. After that, deaths from firearms dropped 59%. Wow, great!

This astonishing statistic left me wondering, so I went looking for more sources, and found the graph below.
Number of homicides per year in Australia.
The red line is added by me, representing the change in policy that occurred in 1996.

Two interesting things about this graph. First, homicides appeared to be on a downward trend even before 1996; less killings were happening each year, and the amount of homicides was within the confidence interval (shown by dotted lines) for the predicted line of best fit through the data. So, it appears Australia was becoming safer each year, even before the 1999 policy.

Second, what happened in 1999 and 2002? The murder rate suddenly spikes, well above the confidence interval for the predicted homicide rate. Apparently the number of deaths from firearms went down 59% overall from 1999-2006, but in 2002, the total number of deaths is considerably higher than any year in the past decade.

The only explanation is that after giving up their guns, Australians found other ways to kill each other that year. After that, homicides continued on their generally downward trend, just as they had before 1996.

The lesson: controlling guns is not a free ticket to preventing murders. Don't underestimate the ingenuity of an evil or unbalanced mind.

3 comments:

  1. "So, it appears Australia was becoming safer each year, even before the 1999 policy." lol based off a 9 year period of observation? that doesn't sound like sufficient data to base that conclusion.

    "The only explanation is that after giving up their guns, Australians found other ways to kill each other that year. After that, homicides continued on their generally downward trend, just as they had before 1999." Or... it took the Australian government 3 years to implement their policy effectively. Or... a substantial amount of gun ownership creates a culture of violence that took 3 years to dissipate. Or... the earlier statistics were poorly measured and adapted a better method for gathering data that took effect in 1999. or like, a grip of other possibilities. The idea that this graph discounts the benefits of Australian gun control efforts seems like a pretty far stretch of the evidence.

    You're loving former debate partner trying to troll your blog because none of your posts are evidence in support of a modernist critique of postmodern aesthetics,

    Abe

    ReplyDelete
    Replies
    1. Hi Abe. Good to hear from you!

      Re: 9 year period of observation. I wish there were more data points to work with. But I'm stuck with what the Australian Institute of Criminology makes available - after a cursory search, I couldn't find much better numbers accessible.

      Re: delay effects. Maybe the policy took time to implement, but because it was a mandatory gun buyback, it seems that would have taken place pretty quickly. One interpretation of the 1999 spike is that criminals decided to have a "last hurrah" and go shoot someone before they sold back their gun... hardly cause for optimism. I also doubt that gun buybacks do much of anything about a culture of violence - that would take decades not years to reverse - so maybe the culture stayed constant but there were less guns out there to deter criminals, so they committed murders they otherwise wouldn't have. As the statistics are based on counting bodies, it seems implausible to me that would have changed much in the time period either.

      This is not a rigorous defense of gun ownership. It's true, that would take much more than this graph. But, from looking at a few academic papers, I found they would often talk about a reduction in "gun-related homicides" as if that was all that mattered, and completely ignore TOTAL homicides. That's why I like this graph so much... It's not conclusive, but certainly is suggestive that criminals can substitute between weapons when guns are harder to access.

      Sorry I can't help with your critique of postmodern aesthetics (does such a thing even exist?) Check back in a couple weeks and maybe I'll have something relevant for you ;)

      Delete
  2. You are exactly correct. It is total unjustified homicides that have to be used in any pragmatic based argument. If total homicides increase or remain the same after a gun ban, then the gun ban was a failure for the purpose of reducing homicides.

    ReplyDelete