(Page 3 of 5)
To illustrate, let the probability (p) that a single individual carries a concealed handgun be .05. Assume further that there are 10 individuals in a public place. Then the probability that at least one of them is armed is about .40 (= 1 - (.95)10). Even if (p) is only .025, the probability that at least one of 10 people will be armed is .22 (= 1 - (.975)10).
Ehrlich claims that I fail to account for all relevant variables. Sure, there could possibly be still other variables out there, though I doubt it. The data used in the first edition of the book have been made available to academics at 45 different universities. I know of no study that has attempted to account for as many factors as I have, but if Ehrlich thinks that other factors are important, he is perfectly free to see whether including them alters the results. Other academics have tried different variables -- for example, Bruce Benson at Florida State University tried including other variables for private responses to crime, and Carl Moody at the College of William & Mary used additional variables to account for law enforcement -- but so far none of these other variables has altered the results.
However, the variable list that I attempted to account for is much more extensive than Ehrlich indicates. Among the factors that I accounted for in the first and second editions of my book are: the execution rate for the death penalty; conviction rates; prison sentence lengths; number of police officers; different types of policing policies (community policing, problem-orientated policing, "broken window" strategies); hiring rules for police; poverty; unemployment; four different measures of income; many different types of gun control and enforcement; cocaine prices; the most detailed demographic information on the different age, sex, and racial breakdowns of the population used in any study; and many other factors.
Discovering some left-out variable is more difficult than simply saying that other factors affect the crime rate. This left-out factor must be changing in the different states at the same time that the right-to-carry laws are being adopted. In addition, crime rates are declining as more permits are issued in a county, so the left-out variable must similarly be changing over time. Other evidence that I presented in my book indicates that just as crime rates are declining in counties with right-to-carry laws, adjacent counties on the other side of state borders in states without these laws are experiencing an increase in violent crime. The more similar these adjacent counties, the larger the spillover. Right-to-carry laws also reduce crime rates where the criminal and the victim come into direct contact with each other relative to those crimes where there is no such contact. To alter the results, these left-out factors would have to vary systematically to coincide with all these different results.
One of the reasons I graphed the before-and-after trends as well as the year-to-year variations in crime rates was to allow readers to judge for themselves whether the adoption of right-to-carry laws coincided with changes in crime rates. For a general audience, I thought that this graphical approach was the most straightforward.
As to the appropriateness of a particular statistical test, the answer depends upon what question one is asking. The one test that Ehrlich questions asked whether there was a statistically significant change in the slopes in crime rates before and after the laws are adopted. For that question, the F-test that I used is the appropriate test.
Research by Florenz Plassman and Nicolaus Tideman that is forthcoming in the October 2001 issue of the Journal of Law and Economics breaks down crime data by each state and by individual years before and after the adoption of the right-to-carry law. They find that for all 10 states that adopted such laws between 1977 and 1992, murder, rape, and robbery rates fell after adoption. If Ehrlich were to identify the statistical test which he says shows a significant turning point for robbery before the adoption of right-to-carry laws, I would be happy to comment on it.
It is flattering that my research is the first topic that Ehrlich discusses in his book, Nine Crazy Ideas in Science. My research, however, is not alone in studying this issue. A large number of academics have examined the data. While a few academic articles have been critical of some of the methodology, not even these critics have found a bad effect from right-to-carry laws. In fact, the vast majority of academics have found benefits as large or larger than the ones I report.
What is also interesting is how little criticism there is of the other gun control topics that my book addressed. For example, no academics have found significant evidence that waiting periods or background checks reduce violent crime rates. Unfortunately, what I have found is that many of these gun control laws actually lead to more crime and more deaths.
In his book, Ehrlich awards "cuckoos" to the ideas he discusses, with one cuckoo meaning "Why not?" and four cuckoos meaning "certainly false." He gives my work three cuckoos, but there are a lot of academics who must then be in the same boat as I am. More important, his criticisms are based upon either an incomplete or inaccurate reading of my work.
Lott's numbers don't tell us anything
I reply below to the main criticisms of John Lott -- at least those which I have understood.
Lott doesn't deny that he misleads the reader by neglecting to mention that his plots are fits to the data, because he can't. His graphs are in fact labeled "number of violent crimes" per 100,000 population and I find no statement in his book that the graphs are fits, rather than actual data. In his reply, Lott justifies the use of displaying fits by noting that it is important to show "adjusted" crime rates after other variables (aside from the laws) have been taken into account.
Lott is correct that I was using the first edition of his book when I made the comment about only 10 states changing their right-to-carry laws in the stipulated time period.