My newest piece at Investors' Business Daily starts this way:
Do fewer guns mean fewer firearm deaths? If you believe the March 6th issue of the Journal of the American Medical Association of Internal Medicine, the answer is "yes."
A study by Eric Fleegler and four other co-authors received massive
national news coverage from USA Today to the television networks.
But the report is based on embarrassingly bad statistics that are rigged to get the result the authors wanted.
Take how they measure gun ownership. Believe it or not, this study
measures gun ownership by looking at the share of suicides committed
using firearms.
Then the authors go on to commit an egregious and basic statistical
error. They claim that states with higher gun ownership have higher gun
death rates. But wait a second — most gun deaths are gun suicides. And
what they call "gun ownership" in their study is also measured by gun
suicides.
In other words, all the study proves is that more gun suicides leads
to, well — more gun suicides. Any serious statistical journal would not
have published such nonsense.
The Fleegler study also involves a geographic comparison across the 50 U.S. states . . . .
The post here completely misread, or more likely did not read, the study methodology. The study is not claiming that owernership increases gun deaths. It is claiming that more gun laws associated with fewer gun deaths. Gun ownership was not used as a MEANS to evaluate the study question, it was used as a correction factor in a multivariate model to correct for possible difference in ownership rates. Without considering ownership rates (models A and B) and considering them (model C), the association between laws and gun deaths was still true.
ReplyDeleteIt's surprising that as an academic you did not recognize this. Perhaps you did not read it in detail yourself.
Sorry, Derek, you should actually read the study. One of their results did deal with gun ownership and firearm deaths.
ReplyDeleteThey also did the do what you say, but I explained in the piece why that is also wrong.