In 1986, Peter Reuter suggested that the Association for Public Policy Analysis and Management (APPAM) consider offering an annual award for the “most outrageous number mentioned in a policy discussion by an elected official or agency head,” with one of the criteria being that the number have “no reasonable basis” (pp. 81 1-8 12).
In this article, we discuss the candidacy of one of the more surprising numbers to surface in the course of America’s gun debate: that 2.5 million Americans use a gun defensively against a criminal attacker each year [Kleck and Gertz, 1995]. News items, editorial writers, even the Congressional Research Service [Bea, 1994] have mentioned the 2.5 million defensive gun uses (DGUs) as established fact. This number is considerably higher than our best estimate of the number of crimes committed each year with a firearm (1.3 million) [U.S. Department of Justice, Bureau of Justice Statistics, 1996b], and has been used as an argument against regulations that would restrict widespread firearms ownership. The implicit notion seems to be that if there are more legitimate uses than criminal uses of guns against people, then widespread gun ownership is a net plus for public safety.
For reasons documented in this article, we believe that the 2.5 million figure is an example of what Max Singer has termed a “mythical number” [Singer, 1971]. Singer notes, “[Elven responsible officials, responsible newspapers, and responsible research groups pick up and pass on as gospel numbers that have no real basis in fact. . . . [B]ecause an estimate has been used widely by a variety of people who should know what they are talking about, one cannot assume that the estimate is even approximately correct” (p. 9).
Estimates for the number of defensive gun uses are likely to be substantially overstated because of the problem of “false positives” [Hemenway, 1996]. This source of bias is a common problem in survey estimates of rare events, but largely unrecognized or ignored. We recount the evidence which indicates that the 2.5 million DGU estimate is far too high, and suggest that implications for both the policy debate over gun regulation, and for survey research.