guns and noise

[Update 5/31: I've attempted to update relevant facts as best I can. While it turns out that yesterday's violence really does make this an exceptionally violent month in Seattle, I think my larger point about statistical noise versus real trends still stands. But feel free to dig into the numbers and draw your own conclusions.]

It's been an unusually violent day here today here in the Emerald City.

Two fatal shootings--unrelated, by all indications--[Ed: SPD now says that the same suspect is believed responsible for both shootings] took place a couple of miles apart from each other within the space of an hour. One was an apparent carjacking, the suspect in which is still apparently on the loose dead of a self-inflicted gunshot wound to the head (see below). The other was a thusfar inexplicable attack on a neighborhood cafe just north of the UW. The latter left two dead at the scene, another two more died later at the hospital just down the block from my office, and two one more are is critically injured. The suspect in the cafe shooting is believed to have killed himself about an hour ago in the middle of the street. [Ed: he actually survived the self-inflicted GSW, was brought to Harborview, and died a few hours later.]

This on the heels of a "spike" in homicides in January and February that have a lot of people worried about violence--and gun violence in particular--in our fair city.

Predictably, these sorts of incidents trigger arguments about guns, gun ownership, and gun control. You may be surprised to learn that while Seattle is indeed a pretty liberal town, Washington state's ownership and concealment regime is about as permissive as any in the country, and no local statute abridges it. Which is to say, the law in Seattle favors broad gun carrying, but the culture tends to frown on it. Kind of an interesting state of affairs.

There are a couple of things worth considering here.

1--Seattle, especially in terms of homicide, is a really safe city, especially compared with other American cities of comparable size. Here are the monthly homicide stats from January 2008 thru March 2012:

A couple of things should jump out at you: in a "typical" month there are 1-3 homicides in the city. In 5 of the 51 months on the graph, there were zero. And in the same time period, there was never a month that exceeded 5. Unless I am missing something (and I may be), the four five homicides today put us at 5 6 for the month of May.

This is a very bad day, making for a bad month. But it is not an unusually bad month. [Ed: I guess it is, now.]

2--Year to date comparisons are not meaningful in this data set. SPD has a practice of comparing the stats of the year to date with the same totals from the same time period the year before. I have no idea why they do this. I suppose it would make sense if there existed some predictable fluctuation of crime over the course of a year. If only we had a way to test this hypothesis...

Same data set as above (minus the first three months of 2012), but this time the data from each month are represented by the box and whisker plot...basically, the line in the middle of the box gives you the average, and the box and whiskers indicate the range of values. If there is a consistent trend over the year, then some months should clearly look different from others. We might be able to infer that Aug-Jan is slightly more violent than Mar-Jun, but this may just be noise. Note that July, the only month for which there is a 5-homicide month in this four-year period, also had at least one zero month. April and September are similarly noisy, ranging from 0 to 4.

There really is no discernible pattern, here.

Why does this matter? Because when we compare January and February of 2012 (4 and 5, respectively) to January and February of 2011 (1 and 1), it looks really bad! It's a 350% increase in the homicide rate! But unless there is a good reason to think that January/February of 2012 ought to look like January/February 2011 (and there isn't, really) the comparison is completely arbitrary. Similarly, when we compare May 2012 (5 and counting) with May 2011 (big, fat zero), May 2012 is literally infinitely worse. But this means exactly nothing.

3--These are really, really small numbers and there aren't very many of them. I don't want to get too deep into statistical theory, here--mostly because I am ill-suited to do so--but suffice it to say that if you want to know whether one set of numbers is statistically significantly different from another, it takes more measurements to determine a smaller difference between the sets, with any reasonable degree of certainty.

I don't want to minimize the horror experienced by victims of violence, here or anywhere, today or any other day. And there may very well be a disturbing trend underway in Seattle. But the fact is that we will not know that until more time goes by.

In the meantime, if someone wants to make an argument that easy access to guns makes gun crime more likely, by all means do so. While I'm personally OK with private gun ownership by responsible, law-abiding citizens, I find many of the pro-gun arguments appalling (in particular, the assertion that there is no relationship between having a gun, and using a gun, and not having a gun, and not using a gun.) I also think some more reasonable allowances ought to be made for local regulation...what's fine and dandy for Cle Elum really may not be appropriate for Capitol Hill.

But...if someone wants to lay the recent violence here on our permissive gun laws, they also have to contend with all of the peaceful months we've had--recently!--under the exact same regime.

(All data shown were compiled into graphs by me, based on numbers obtained from SPD. For some reason, SPD only published annual totals in 2010, so I got my numbers for 2010 from this Seattle PI piece. I cannot attest to their accuracy.)

No comments: