A few days ago, David Kopel posted on Volokh about a new paper he and coauthor Howard Nemerov have written. The subject is the relationship between firearm ownership and "freedom and prosperity," the latter of which is measured through some widely-known indicators (Freedom House scores, Transparency International corruption ratings, World Bank prosperity scores, and the Heritage Foundation's economic freedom index).
To their credit, the authors are extremely careful in their conclusions. Examining data for 59 countries during the mid-2000s, the authors
conclude that "(T)he highest national levels of gun ownership are
associated with more political freedom, more civil liberty, much less
corruption, and slightly more economic freedom." They also note, however, that "the relationship is not universal," and go to pains to discuss the many ways in which their data -- particularly that on gun ownership -- may have problems. Perhaps most interesting is their statement, based on a series of comparisons of means across groups defined by quartiles of gun ownership, that
(W)hile the first quartile is better in all categories, the relationship between firearms and freedom is not consistent among the lower three quartiles. For example, the second quartile is slightly better for economic freedom, the third quartile is best for non-corruption, and the fourth quartile is best for political/civil freedom. Thus, if there is some kind of cause-and-effect relationship between firearms and freedom (discussed infra), the effect appears to exist only for countries with high levels of firearms ownership. The effect does not appear evident between groups of countries with relatively low levels of firearms vs. countries with hardly any firearms.
Of course, the analysis is far from perfect, and several of the commenters at Volokh offer suggestions about how the authors might improve their analysis. One suggestion -- with which I agree -- is to provide some graphs of the relationships among the variables they consider. Since the authors are kind enough to make their data available in the paper, I decided to do just that:
The variables are all rescaled so that higher values are "better." Green dots are countries Freedom House defines as "free," black ones are "partially free," and red ones are "not free;" dotted lines are linear fits, smooth ones are lowess lines, and the main diagonal has density plots for each variable.
The key points are the positive relationships among freedom, transparency, etc., the somewhat weaker relationships between those indicators and gun ownership, and -- interestingly -- the strongest evidence for the authors' "step function" idea in the relationship between firearm ownership and transparency. Most important, though, is the fact that, particularly when it comes to data analysis, a picture really is worth a thousand (or more) words.
Recent Comments