John Lott responds to criticism of his study that appeared in the Economist with a letter that was sent to the magazine, some additional comments, and a response to a question from GunCite. (Quoted material from John Lott is from correspondence with GunCite on January 20, 2001.)

Dear Letters Editor:

Your article linking gun magazine sales to violent crime rates makes several mistakes and mischaracterizes the academic debate (Do guns mean crime?, January 13).

The Economist is simply wrong to claim that my research used concealed handgun laws as a proxy for gun ownership. Although it is likely that more people will buy guns, concealed handgun laws can reduce crime even if gun ownership remains constant because people can use guns to protect themselves in places where they previously went undefended.

It remains a mystery why Mark Duggan's paper relies on the current sales of only the fourth largest gun magazine (Guns & Ammo) as a proxy for gun ownership when state-by-state survey data is available. My book (More Guns, Less Crime) used survey data to measure gun ownership and found that those states with the biggest increases in gun ownership had the biggest drops in violent crime.

It is true that Duggan found a "relationship between the level of magazine sales and the number of guns shows in states," but the Economist fails to note that, even if gun shows are somehow a strong proxy for gun ownership, magazine sales explained only 15 percent of the variation in the number of shows.

Duggan and the Economist are also wrong to claim that the "particular magazine [Duggan picks] concentrates on handguns," though this point is important to their argument. Over the entire 1980 to 1998 period, 50 percent of the gun reviews are on long guns. If one really wanted to focus on handgun sales, the very company that publishes "Guns & Ammo" also produces the magazine "Handguns." Other handgun magazines are available.

As to the research by Hashem Dezhbakhsh and Paul Rubin, the Economist apparently does not realize that they neither examined whether concealed handgun laws explained why crime rates were falling nor used the same data that I did. Their paper did not investigate how crime rates changed before and after concealed handgun laws were passed and only used 20 percent of the data I used.

While there have been a few papers critical of my research, the debate has been over how large the benefits are, with some claiming that the benefits are small or zero and many more claiming to find large drops in crime, some even larger than I did.

Sincerely,

John R. Lott, Jr.
Senior Research Scholar
School of Law
Yale University
New Haven, CT 06511

Additional comments from John Lott:

1) Guns & Ammo is a relatively small gun magazine, amounting to 16% of sales of just the top four gun magazines.

2) The demographics of their subscribers are not similar to what the GSS or the VNS indicates most guns owners look like either in terms of sex or age and I think also race. 79.5 percent of subscribers go "Big Game" hunting. 80 percent own a light truck. 84.4 percent have bought boots for hunting within the last two years and virtually, literally everyone (99.7 percent) owns at least one dog. Other than this variable tracking young males (the typical subscriber), I am not sure what relationship one should expect that this variable should have with crime.

3) Even if one thinks that magazine sales are the right measure of gun ownership, initial work indicates that the results do not hold true for any other gun magazines.

GunCite asked Lott to respond to the following criticsm from Mark Duggan's paper:

One problem with these regression estimates is that Lott and Mustard are implicitly assuming that these laws are varying at the county level, when in fact they are varying only at the state level. [Footnote excepting Philadelphia, omitted] Therefore, one must adjust the standard errors appropriately to account for the fact that county-level disturbances are likely to be correlated within a state. The second column presents coefficient estimates for p from nine analogous regressions that properly account for the within-state correlation in errors. In all cases, the standard errors increase more than threefold, and only the auto theft estimate remains statistically significant. Thus it appears that the passage of CCW legislation is not associated with a significant decline in rates of violent crime, though all four violent crime coefficent estimates are negative.

Lott's response:

The correlation of the error terms across counties is picked up when one has county fixed effects included in the regression. He does not do the adjustment recognizing that the county fixed effects are already picking up what he wants to adjust for.