Tapsearch Com EditorNew pro-gun control study out
January 13th 2007 19:06
Alphecca has extensive coverage of a new study on gun control. The site critiques the study on several grounds, most interestingly pointing out it was funded by an anti-fun (oops, gun, Freudian slip) organization.
From the Harvard press release:
"Analyses that controlled for several measures of resource deprivation, urbanization, aggravated assault, robbery, unemployment, and alcohol consumption found that states with higher rates of household firearm ownership had significantly higher homicide victimization rates for children, and for women and men. In these analyses, states within the highest quartile of firearm prevalence had firearm homicide rates 114% higher than states within the lowest quartile of firearm prevalence. Overall homicide rates were 60% higher."
My reaction to all this is, so what? It's a basic correlation-causation problem. There's no way to tell whether the guns cause the crime, or whether people buy guns to protect themselves from the crime.
But:
"The association between firearm prevalence and homicide was driven by gun-related homicide rates; non-gun-related homicide rates were not significantly associated with rates of firearm ownership."
If people were buying guns in response to crime, you'd expect gun ownership rates to be associated with crime regardless of whether the crime was gun-related.
Here's the most important, thing, though. As Alpecca cites from this story:
"After dividing the states into four groups based on how many households had guns, the researchers found the states in the highest quartile of firearm ownership had overall homicide rates 60 per cent higher than states in the lowest quartile."
Anyone with basic knowledge of statistics (I have little beyond that) would ask: Why divide the states into four groups? There's no reason a competent mathematician with access to Excel (or even Google Spreadsheets) couldn't run all the controls, regressions and correlations on all the individual pieces of data. That's what John Lott did for his study on gun control -- at the county level, even, not the state one. Incredibly powerful computers have made shortcuts like this unnecessary.
Alphecca alleges that the study also lumps suicides and murders together, a grave flaw if true, but the available information on the study always says "homicide," not "firearm death."
I also love this conclusion from the abstract:
"Although causal inference is not warranted on the basis of the present study alone, our findings suggest that the household may be an important source of firearms used to kill men, women and children in the United States."
Well, duh. The DOJ has documented that, in 1991, "among those inmates who possessed a handgun, 9% had acquired it through theft, and 28% had acquired it through an illegal market such as a drug dealer or fence [who presumably stole the guns themselves]."
Robert VerBruggen blogs at http://www.therationale.com.
UPDATE: Pro-gun scholar John Lott weighs in extensively here with lots of data.
Lott's main contention:
"The bottom line is that their results comes from two factors: the exclusion of DC and the use of other crime rates to explain the murder rate. Changing these two factors causes their result to go from positive and significant to negative and significant."
From the Harvard press release:
"Analyses that controlled for several measures of resource deprivation, urbanization, aggravated assault, robbery, unemployment, and alcohol consumption found that states with higher rates of household firearm ownership had significantly higher homicide victimization rates for children, and for women and men. In these analyses, states within the highest quartile of firearm prevalence had firearm homicide rates 114% higher than states within the lowest quartile of firearm prevalence. Overall homicide rates were 60% higher."
My reaction to all this is, so what? It's a basic correlation-causation problem. There's no way to tell whether the guns cause the crime, or whether people buy guns to protect themselves from the crime.
But:
"The association between firearm prevalence and homicide was driven by gun-related homicide rates; non-gun-related homicide rates were not significantly associated with rates of firearm ownership."
If people were buying guns in response to crime, you'd expect gun ownership rates to be associated with crime regardless of whether the crime was gun-related.
Here's the most important, thing, though. As Alpecca cites from this story:
"After dividing the states into four groups based on how many households had guns, the researchers found the states in the highest quartile of firearm ownership had overall homicide rates 60 per cent higher than states in the lowest quartile."
Anyone with basic knowledge of statistics (I have little beyond that) would ask: Why divide the states into four groups? There's no reason a competent mathematician with access to Excel (or even Google Spreadsheets) couldn't run all the controls, regressions and correlations on all the individual pieces of data. That's what John Lott did for his study on gun control -- at the county level, even, not the state one. Incredibly powerful computers have made shortcuts like this unnecessary.
Alphecca alleges that the study also lumps suicides and murders together, a grave flaw if true, but the available information on the study always says "homicide," not "firearm death."
I also love this conclusion from the abstract:
"Although causal inference is not warranted on the basis of the present study alone, our findings suggest that the household may be an important source of firearms used to kill men, women and children in the United States."
Well, duh. The DOJ has documented that, in 1991, "among those inmates who possessed a handgun, 9% had acquired it through theft, and 28% had acquired it through an illegal market such as a drug dealer or fence [who presumably stole the guns themselves]."
Robert VerBruggen blogs at http://www.therationale.com.
UPDATE: Pro-gun scholar John Lott weighs in extensively here with lots of data.
Lott's main contention:
"The bottom line is that their results comes from two factors: the exclusion of DC and the use of other crime rates to explain the murder rate. Changing these two factors causes their result to go from positive and significant to negative and significant."
| 53 |
| Vote |
Subscribe to this blog
