Thursday, March 17, 2011

Some Thoughts on Statistics and Guns

In the last little while, I have continued a debate with James Kelly.  This is Kelly's third post on a debate that was supposed to die with the first post.  I have committed myself to not rebutting his post, but doing so has annoyed me:  the points really deserve rebutting!  But both of us have already put too much energy in the debate, and it's becoming quite circular.  Even so, I wanted to get a few things off my chest, and to half-rebut several things.  Hence, this post.

As a word of caution:  I'm not in the mood to look up these statistics, so feel free to correct me, or even to provide collaborating links.

First, I found it amusing that Kelly continue to give me statistical reasons to fear guns, when I've stated time and time again that I don't trust statistics--not even my own!  He gave examples of the United States, or Finland, and of Brazil as positive correlations between guns in society and violence, and Great Britain as an example where they had less guns and less violence--then said that my conclusion that there is no correlation between guns and violence is false.  It's interesting that he left off Switzerland (high-gun but low-crime) and former Eastern Bloc countries (low-gun but high crime) from the list of examples.

Second, Kelly never understood my desire to know the "guns per gun death" statistic.  I wanted to test this basic idea:  which society handles guns more safely?  It is my suspicion that the Arizona would a lower rate of deaths per gun than Scotland--which would run counter to the idea that "more guns would be more dangerous in society".  Alas, such data is difficult to find.  Since I'm a big believer in looking at how things change over time, I'm hopelessly lost, because it seems rather difficult to find data like that kept over time, for Scotland.

But I would admit:  I'm not sure if I'd trust such a number, even if it came out in my "favor".  This is because it would rely on the number of illegal guns, which would be, at best, an educated guess.  Heck, the number of legal guns in Arizona would be, at best, an educated guess, because Americans aren't required to register their guns to be legal!

Even so, now that I've been thinking about this, I'm curious:  what is this statistic, and how will it change over time?  I may distrust statistics, but I've learned enough in the course of becoming a mathematician that I have a tiny statistician in my soul that just itches to grab hold of data and run with it.

Third, in the process of trying to find these numbers, I stumbled onto two curious headlines.  One was, in effect, "Scotland's murder is lowest in 31 years!" and the other was, in effect, "The Number of Guns in Scotland Have Increased!"  As I thought about this, I found it amusing:  a correlation that I could have exploited!  I'm not sure anything would have come of it, though.  It's just an amusing coincidence, after all.

Or is it?  That's one of the funny things about trying to use statistics to understand society.  As much as I admire John Lott's work, or Gary Kleck's, or other criminoligists and statisticians, we're trying to measure effects in a very chaotic system--a society of individuals, each with their own free will, making countless decisions over time on how best to act.  It is very difficult, perhaps even meaningless, to point to a single thing and say "This caused that!".

I like to see what, in Calculus, are called "deltas"--changes over time.  It would be impossible to see all the deltas--for example, why did Scots suddenly feel the need to buy more guns?--but if you're going to claim something like "banning handguns will reduce crime", then, to convince me of this, you'd better show me a reduction in crime after you ban handguns.  Unfortunately, in Great Britain, violence has increased since the handgun ban.  Kelly dismissed this line of reasoning, saying "handguns were practically banned before the formal ban, yet crime was rising before!"  Well, his exact words were,
The whole concept of householders routinely owning or using guns for 'defensive' purposes was already an alien one prior to 1996. A practice has to have meaningfully existed before its 'removal' can be claimed to have made a difference.
which is odd, because if the idea of having legal access to guns increases violence, rather than decrease it, then one would think that crime should have been decreasing well before 1996.  Yet here's an essay (the only one I'll link to) that discusses how murder rates decreased steadily over six centuries as technological advances allowed more people in Great Britain to own guns, only to increase, beginning in 1920, when Great Britain began to ban guns.

Admittedly, it's based on a study--which means it's contestable--but if I'm to be convinced that gun control works, I'll first need to see studies that show the reverse, and have even more solid footing than the one to be contested.

So, there you go:  some thoughts on statistics, based on a long and convoluted argument.

8 comments:

  1. This is something of an interesting twist for James, given that Kevin Baker beat him half senseless with reams after reams of statistics, data, and information, and James dismissed it all as completely meaningless and irrelevant. Now he is using statistics of his own to support his personal phobias? Hm. I smell freshly turned earth and repositioned goalposts...

    I, too, would be interested in "firearm-related deaths per firearm" statistics - such numbers are relatively easy to come by on a national level (allowing for certain assumptions), but on a state-by-state... bleh, basically impossible.

    But, as you say in your second post, it does not really matter what the statistics say - our rights exist independent of them, no matter how they may be interpreted or used.

    ReplyDelete
  2. "beat him half senseless"

    What is it with you and your rape fantasies, Linoge? I made no secret of the fact that I changed tack last year - I was repeatedly pestered by your side of the argument to engage on the statistical evidence, so eventually I gave in and did what you claimed to want, with fascinating results. Remember my ten-question challenge? It took weeks before anyone was prepared to attempt a comprehensive answer. Kevin himself only addressed one of the ten questions, and not terribly convincingly.

    Epsilon, I've discussed Swizerland in the past - its gun death rate is significantly higher than the UK's.

    "One was, in effect, "Scotland's murder is lowest in 31 years!" and the other was, in effect, "The Number of Guns in Scotland Have Increased!" As I thought about this, I found it amusing: a correlation that I could have exploited!"

    Well, that cuts both ways, doesn't it? If, more than a decade after the handgun ban, Scotland has just seen its lowest murder rate in 31 years, that does rather undermine the theory that strict gun control regimes inevitably put people at more risk of harm (ie. as a result of "leaving only the criminals armed").

    I didn't misunderstand your interest in the "gun per death" statistic - I just don't think it's relevant, for the reasons I've explained a couple of times before.

    ReplyDelete
  3. I'm sorry, James, but I already told you why statistics don't cut it for me. Statistical evidence made me curious about guns, but it was the philosophy that brought me into the culture--not just of the culture of guns, but of freedom in general.

    I have not seen your ten questions, and if they are statistical in nature, I will probably just shrug them off anyway.

    Over the years, I've seen too many studies, both for and against, on this issue. As a mathematician, I have an "inner statistician" that is curious about statistics, and I certainly have interest in gathering data, and in number crunching--but I have gone well beyond being persuaded by statistics. Every so often, I'll look at one and say "see, told you so!".

    It is ultimately ironic that you are harping on statistics, when you yourself stated that you have no interest in statistics!

    ReplyDelete
  4. Well, that cuts both ways, doesn't it? If, more than a decade after the handgun ban, Scotland has just seen its lowest murder rate in 31 years, that does rather undermine the theory that strict gun control regimes inevitably put people at more risk of harm (ie. as a result of "leaving only the criminals armed").

    Well, yes it does, sortof, except that, since this year came after a decade of increase in violent crime, we can legitimately suspect that other forces may very well be at work. That's the funny thing about statistics.

    Furthermore, as I stated earlier, the murder numbers are so darn small, I ultimately have serious questions about the legitimacy of any claim that banning guns will result in less murders, or less violence.

    ReplyDelete
  5. *sigh* And, predictably, James is now resorting to making up nonsense out of whole cloth and incessant (and ironic, as you mentioned) goalpost shifting.

    Suffice to say, there are more than a few reasons most people did not bother with your ten questions, and "because they could not formulate a good answer" is somewhere near the bottom.

    ReplyDelete
  6. "I have not seen your ten questions, and if they are statistical in nature, I will probably just shrug them off anyway."

    The questions actually challenged the credibility of statistical evidence that the pro-gun side had advanced.

    As for "irony", I can't think of anything much more ironic than Linoge moaning about me changing tack last year and engaging on the statistical front, when he and others repeatedly told me that my philosophical arguments would have no credibility until I did so!

    ReplyDelete
  7. ... And again with the nonsense. Remember that list I was talking about? Thiswant them to say.

    Rather defeats the purpose, does it not?

    ReplyDelete
  8. HTML tag fail.

    As I was saying, "This is somewhere near the top - your apparent inability or unwillingness to respond to what people actually say rather than what you want them to say."

    ReplyDelete