The graph at the link exposes one of the major problems with small--or short timeline--data sets.
If we take out the two obvious outliers, the trend line gets a whole lot flatter.
That raises a couple of questions: in what ways are those two outliers important? What, if anything, is unique to them and sets them so much apart--is there a set of criteria separate from the general that generated them? Or, is there a common set of criteria that happened to converge in those two years (if so, what and why)? Etc.
Using the graph by itself as a serious argument, without addressing those kinds of questions, reduces the credibility of the less-gun-control argument.
Also from the linked-to article: Fears of teachers carrying guns in terms of such problems as students obtaining teachers guns have not occurred at all....
Overemphasizing this sort of thing also puts at risk our argument. No system is perfect; some students, some time, will get hold of a teacher's gun, and a subset of those students will commit mayhem with the gun. The better argument is to emphasize the unlikelihood of such a thing, especially compared with the rate of guns brought in from outside for mayhem committing.
If we take out the two obvious outliers, the trend line gets a whole lot flatter.
Sure. And if we assume that tomorrow there will be a big shooting in a school with armed teachers, the other trendline becomes a lot less flat (in fact, it becomes a sudden spike). But this is what the data show, for now.
Statistics aren't a very useful tool in cases like this, where events are very rare. As Lott's paper points out, the rate is zero for schools that allow students to carry guns; but it's only 0.039 per 100,000 for the schools without armed teachers. At that level of rarity, one event can throw the picture off wildly.
I'm sorry, but the information presented is useless. It isn't scaled by either the number of pupils in "gun-free/non-gun-free" schools, or by the number of such schools. The paper throws around a few numbers, but doesn't support them or commit to anything solid. The excel spreadsheet with the data makes interesting reading, but without the population size information I don't know anything about the rates.
It is also important to compare apples to apples with respect to demographics: schools with the same socio-ethnic background with and without the gun policy. That will undoubtedly make the statistics even worse, since this is plainly a rare occurrence.
You could expand the study to other kinds of violence, though at the expense of making it clearly gun-policy related.
Let me put it this way--I propose an alternative hypothesis: The schools in areas open to allowing teachers to carry are areas with less violence to begin with. The areas with more trouble tend to be those with more government control of life (I don't propose to argue cause/effect here, but I think they reinforce each other), and therefore less likely to allow teachers to arm.
Can you distinguish my hypothesis from his with that data?
As those are raw numbers, and the population of children in our nations primary and secondary schools numbers well over 60 million, I have a slightly different takeaway- That we're really discussing something on the level of getting struck by lighting (if even that common). It's a bit ridiculous to have all this debate and policy setting over something that's in reality so very rare.
The whole discussion should be dropped from at least the Federal level.
The graph at the link exposes one of the major problems with small--or short timeline--data sets.
ReplyDeleteIf we take out the two obvious outliers, the trend line gets a whole lot flatter.
That raises a couple of questions: in what ways are those two outliers important? What, if anything, is unique to them and sets them so much apart--is there a set of criteria separate from the general that generated them? Or, is there a common set of criteria that happened to converge in those two years (if so, what and why)? Etc.
Using the graph by itself as a serious argument, without addressing those kinds of questions, reduces the credibility of the less-gun-control argument.
Also from the linked-to article: Fears of teachers carrying guns in terms of such problems as students obtaining teachers guns have not occurred at all....
Overemphasizing this sort of thing also puts at risk our argument. No system is perfect; some students, some time, will get hold of a teacher's gun, and a subset of those students will commit mayhem with the gun. The better argument is to emphasize the unlikelihood of such a thing, especially compared with the rate of guns brought in from outside for mayhem committing.
Eric Hines
If we take out the two obvious outliers, the trend line gets a whole lot flatter.
ReplyDeleteSure. And if we assume that tomorrow there will be a big shooting in a school with armed teachers, the other trendline becomes a lot less flat (in fact, it becomes a sudden spike). But this is what the data show, for now.
Statistics aren't a very useful tool in cases like this, where events are very rare. As Lott's paper points out, the rate is zero for schools that allow students to carry guns; but it's only 0.039 per 100,000 for the schools without armed teachers. At that level of rarity, one event can throw the picture off wildly.
I'm sorry, but the information presented is useless. It isn't scaled by either the number of pupils in "gun-free/non-gun-free" schools, or by the number of such schools. The paper throws around a few numbers, but doesn't support them or commit to anything solid. The excel spreadsheet with the data makes interesting reading, but without the population size information I don't know anything about the rates.
ReplyDeleteIt is also important to compare apples to apples with respect to demographics: schools with the same socio-ethnic background with and without the gun policy. That will undoubtedly make the statistics even worse, since this is plainly a rare occurrence.
You could expand the study to other kinds of violence, though at the expense of making it clearly gun-policy related.
Let me put it this way--I propose an alternative hypothesis: The schools in areas open to allowing teachers to carry are areas with less violence to begin with. The areas with more trouble tend to be those with more government control of life (I don't propose to argue cause/effect here, but I think they reinforce each other), and therefore less likely to allow teachers to arm.
Can you distinguish my hypothesis from his with that data?
As those are raw numbers, and the population of children in our nations primary and secondary schools numbers well over 60 million, I have a slightly different takeaway- That we're really discussing something on the level of getting struck by lighting (if even that common). It's a bit ridiculous to have all this debate and policy setting over something that's in reality so very rare.
ReplyDeleteThe whole discussion should be dropped from at least the Federal level.