The Latest Misfires in Support of the More Guns, Less Crime Hypothesis

Abstract:     
John Lott, Florenz Plassman, and John Whitley ("LPW") have criticized our article, Shooting Down the More Guns, Less Crime Hypothesis, by arguing that some aggregated statistical models that we criticized support their "more guns, less crime" claim (which leads them to say we "misread" our results) and by offering new regressions on an expanded county data set. We maintain, however, as we did in our original article, that the aggregated models favored by LPW are flawed by a serious selection effect problem (and in any event we show that the findings LPW point to are undermined by controls for pre-existing state trends in crime). Indeed, we illustrate that simply dropping the states that adopted concealed carry laws during the crack epidemic leads to estimates that concealed carry laws strongly increase crime (which underscores the importance of the omitted crack phenomenon in driving the initial Lott and Mustard results). Moreover, we discovered that the ostensibly supportive results obtained by LPW after extending their county set to 2000 were caused by some mis-coding errors they made in extending their data. When we correct these errors, their findings are reversed: LPW's preferred spline model fails to generate a statistically significant effect for any crime category, while the only significant results in the other possible models show the laws to be associated with increases in various property crimes (and in one case for rape).

Working Paper Series