GUNNING DOWN CRIME
The Statistics of Concealed Weapons
Jun 1, 1998, Vol. 3, No. 37 • By NELSON LUND
For thirty years, crime and guns have been among the most hotly disputed and politically potent issues in America. Not surprisingly, slogans and anecdotes have been the principal ammunition. When empirical evidence has been allowed at all into debates about gun control, it has been notable mostly for its crudity. Proponents of gun control never tire of praising the low murder rates in Japan and Great Britain (which tightly restrict civilian access to firearms), while their opponents are equally tireless in pointing out the low murder rates in Switzerland and Israel (where gun ownership is widespread).
With few exceptions, academic studies of gun control have barely exceeded the quality of the sound bites fired off by warring interest groups. John R. Lott Jr., an economist who specializes in measuring the effects of legal rules, has now published the results of empirical research that is much more detailed and far more sophisticated. Indeed, Lott has gone so far beyond other scholars that his work deserves a central place both in future academic inquiry and in popular and political debate. Unfortunately, Lott's book may not get its due.
As the title suggests, More Guns, Less Crime concludes that the proliferation of firearms among the civilian population has good effects. Lott shows that giving ordinary citizens the freedom to carry concealed weapons in public is followed by reductions in such crimes as murder, rape, and robbery. This conclusion should come as no surprise, though the media reacted to its initial announcement as though Lott had claimed to find an exception to the law of gravity. But what does seem surprising is how dramatically crime rates fall when laws regulating concealed weapons are liberalized. This finding has profound implications for public policy, and the real importance of Lott's book lies in the extraordinary care and objectivity with which he comes to his conclusions.
The thesis that guns lessen crime is plausible, but not self-evident. And even if it is true, legislators and policy analysts need to know how much harm is done by laws that forbid the carrying of arms. Lott's research, which applies sophisticated statistical analysis to an unusually large and high-quality set of data, provides estimates of the magnitude of the harm caused by certain forms of gun control.
In evaluating Lott's study of the correlation between changes in gun laws and changes in crime rates -- and his conclusion that liberalizing concealed-carry laws has strong deterrent effects on violent crimes -- it is important to note that no such study can ever be definitive. The first reason is that such a correlation can never conclusively establish causation: Even a perfect correlation between roosters crowing and the sun rising wouldn't prove that roosters cause the sun to rise -- just as it wouldn't prove that the rising sun causes roosters to crow, for the data leave open the possibility that other factors are causing both the crowing and the rising. Statistical relations are informative only to the extent that they tend to confirm or falsify a theory of causation.
The second reason such studies can never be definitive is that perfect data sets do not exist. Testing a new drug on a hundred patients for five years might indicate whether the drug lacks significant adverse side effects, but a study of a thousand patients over twenty years would necessarily produce a more reliable indication, and a thirty-year study of a hundred thousand patients would be better still.
And yet, although such research can never claim finality, Lott's results are unusually powerful. The theory of causation that he tests is simple and plausible: Making crime riskier reduces the likelihood that crimes will be committed. Just as we would expect burglar alarms and guard dogs to reduce the incidence of burglaries, so should we expect an increase in the number of potential victims who are armed to reduce the incidence of violent crimes. So too, Lott's data cover more than three thousand jurisdictions over a period of eighteen years, which provides him with many more observations than used in previous studies. And finally, Lott has been extremely industrious, imaginative, and scrupulous in performing statistical tests designed to control for factors other than changes in concealed-carry laws that might influence crime rates.