bookmoth

This is an analysis of an article that's been floating around the gunblogs these days. for more, see SayUncle, geekwitha.45, and John Lott.

I turned this in for a Master's level class in scientific methodology, so I don't (think I) swear.

In Press: Hemenway, D., Vriniotis, M., Miller, M. Is an armed society a polite society? Guns and road rage.” Accident Analysis and Prevention.

This study attempts to empirically analyze the statement “an armed society is a polite society” with a telephone survey looking at aggressive driving and gun ownership / possession. Immediate problems arise in the introduction. As the authors point out, communication between cars is difficult at best. Directions, apologies, and minor threats are difficult. Road rage, as the authors use the term, is visibly aggressive driving that can involve illegal actions (tailgating or brandishing a gun, for example). The authors claim that the “armed society is a polite society” quote yields 33,000 websites from an Internet search then claim “no one seems to have explained what the phrase means” (1), a statement that implies they read all 33,000 websites searching for the meaning and that they read the source text, Beyond This Horizon by Robert Heinlein.

So the study operationalizes “armed” to mean owning a gun or carrying a gun in one’s car, loaded or unloaded. “Polite” is operationalized to mean not engaging in aggressive behavior while driving. I question whether a person can be called “armed” when driving with an unloaded gun locked in the trunk with ammunition locked in another container, and I question whether “polite” is simply the absence of illegal, aggressive behavior.

The study is a survey of 2459 people conducted via random telephone calls. Of the calls that were successful (when the experimenters spoke to an adult of the intended, randomized gender, etc.), 31% of potential subjects refused to participate. From what I know of survey response, this is a high response rate, but I believe it is biased toward people who do not own guns, people who own guns illegally, and people who are irresponsible gun owners. My own (limited) experience with responsible gun owners has shown me that they are extremely cautious about revealing themselves to own guns, especially to people perceived as “anti-gun” (the experimenters are from Harvard and received money from the Joyce Foundation); people change pediatricians over this issue and decide where to live based on gun laws. At least some of the gun questions in the survey were worded improperly. For example, the question, “…how many days were you in a motor vehicle in which there was a gun” could lead to false positives when correlated with questions like “…have you been arrested…” because one can be in a car but not own or be in possession of the firearm. The “gun-in-car” question was turned into a yes/no (anything over 0 = yes). The authors also asked if the subject owned any firearms that were in their home. The study focused on an analysis of these four categories.

6.3% of the subjects who admitted making obscene gestures and following aggressively carried a gun in their vehicle at least once in the past year, compared with 2.8% of subjects who never carry a gun (p<0.001), with a multivariate odds ration of 1.7 (95% CI = 1.0, 2.9). Other results showed that male gender, smoking, binge drinking, youth, and a criminal record were also predictive of aggressive driving. The authors note briefly that those who carried guns were both more likely to report being the victim of impolite behavior.

The questions asked do not look at complete incidents (was a reported rude gesture a response or instigation? Was the subject armed during any of the incidents? Did the subject have access to the firearm and was the firearm operational?). The authors pad their discussion section with a few comments that point out these shortcomings. They reference studies defining what makes a good driver and studies that show correlation between driving drunk, smoking, youth, living in urban areas, and being male and aggressive driving. Obviously correlation does not mean causation. The authors do not go so far as to claim that it is, but they do say “riding with a firearm in the vehicle appears to be a marker for aggressive and dangerous driver behavior”(8). But they don’t quantify the correlation. An increase from 2.8% to 6.3% is more than doubling, but the figures are still relatively small, and the authors don’t mention the strength of the correlation compared to other mentioned factors.

Finally, the authors acknowledge funding from the Joyce Foundation, an anti-gun nonprofit. This raises the issue of bias.

In sum, I give them a D-. I would fail them, except they had a lot of statistics and made some interesting general observations. But their methodology, conclusions, and impartiality are all flawed.