Myth #3 - "2.5 million defensive gun uses each year can't be accurate"

Myth #3 - "2.5 million defensive gun uses each year can't be accurate."

In a nation of some 97 million households, about half of which own one or more firearms, is a rate of defensive gun use (DGU) which amounts to about 1.4% per year an unreasonably high figure?

A study undertaken by a group led by criminologist Dr. Gary Kleck of Florida State University found that there are approximately 2.1 to 2.5 million instances annually in which individual Americans use a gun to defend themselves. Considered as households, the figure is 1.3 to 1.5 million annual DGUs (Kleck 1995, Table 2). If this figure is correct, defensive uses of firearms are much more common than crimes committed with guns. Kleck's study defines a DGU as a defensive action against a human (rather than an animal), involving actual contact with the person being defended against, in which the defender could state a specific crime which he or she thought was being committed at the time of the incident, and in which the defender's gun was actually used in some way, even if it was only as part of a verbal threat. A reported DGU incident must meet all of these criteria in order to be counted as a valid DGU for the purposes of the survey. Additionally, DGUs associated with work as a policeman, security guard, or member of the military are excluded.

The data for the Kleck study were collected using an anonymous nationwide random-digit-dialed telephone survey of 4,977 adults conducted from February through April of 1993 by Research Network, a telephone polling company located in Tallahassee, FL. After a few general questions about problems in their community and crime, those polled were asked "Within the past five years, have you yourself or another member of your household used a gun, even if it was not fired, for self-protection or for the protection of property at home, work, or elsewhere? Please do not include military service, police work, or work as a security guard." Those who answered "Yes" were then asked whether their defensive use was against an animal or a person, asked to state how many defensive gun use incidents against persons had happened to members of their household in the last five years, and asked whether any of the incident or incidents had occurred in the last twelve months. Of those surveyed, 222 respondents reported DGUs within the past five years. All respondents reporting DGU, as well as 20% of those not reporting a DGU, were called back to validate their initial survey interviews. These raw data were then corrected for oversampling in the South and West regions, where gun ownership is highest; and oversampling for males, who are not only more likely to own guns, but also more likely to be victims of violent crime.

The weighted results (corrected for oversampling built into the survey) were these: 1.125% to 1.326% of respondents reported having personally been involved in a DGU incident within the past year, with 1.366% to 1.587% of households reporting a household member being involved in a DGU incident within the past year (which would include those DGUs mentioned above involving the respondent). Calculations based on the estimated adult population of the U.S. and the estimated number of households in the U.S. show that at this rate there would be 2,163,519 to 2,549,862 DGUs in 1993 if considered on an individual basis, or some 1,325,918 to 1,540,405 DGU-involved households. For comparison, the estimated number of violent crimes committed with guns in 1993 was 588,140, according to the FBI's Uniform Crime Reports.

Kleck discusses the flaws inherent in previous surveys, including the Justice Department's National Crime Victimization Survey (NCVS), which gives the lowest estimate of all methods used to calculate the number of DGU incidents. The NCVS, Kleck argues, though based upon a much larger sample than his own survey, severely undercounts the number of DGUs because respondents are being asked by a government official to volunteer information about incidents in which they may possibly have done something that was illegal (such as committing assault with a deadly weapon by pointing a gun at someone), or which involved an illegal act (such as carrying a gun without a permit). Further, the NCVS never asks directly whether the respondents used a gun to protect themselves, and only asks its general question about self-protection after respondents have already reported the location of their victimization incident, which in most cases is reported as being away from the victim's home. Since carrying a gun without a permit is often illegal, there would be a strong motivation for respondents not to report DGU outside the home. But how accurate is the Kleck survey's estimate? What are some possible alternative explanations that could influence these results?

First, consider some factors which could produce an overestimate. The sample size, or number of persons surveyed, while smaller than the NCVS, is several times larger than that usually used in most public opinion polls. Extrapolating from too small of a sample is one way to get an overestimate, but this is not a problem for Kleck's survey, since the Gallup Poll and the Roper Center routinely use samples smaller than this for their national surveys. (The results of such national polls are often promoted as newsworthy.) As Kleck notes, the random sampling error of his survey is less than 1%. The phenomenon known as "telescoping," in which past events are remembered as being much more recent than they actually are, could account for some increase in reported DGUs, but Kleck's survey, which uses two time periods, asking about events in the past five years, and within the last twelve months, takes this into account. Kleck's definition of a DGU to include defense of property could serve to increase the number of reported DGUs, since use of deadly force to protect property is not legally recognized in many states. Kleck notes however, that it was not the purpose of the survey to discern the legality or morality of the respondent's actions.

One possible source of overestimation that has been seized upon by critics wanting to explain away Kleck's results is what Kleck terms the "dishonest respondent" hypothesis, or in other words, that the respondents made up a story about a DGU. Only 24% of respondents reporting a DGU claimed to have fired their gun, and even if one were to discard these reported DGUs entirely (and arbitrarily) as overly dramatic, this leaves well over a million and a half estimated DGUs, a rate twenty times that of the estimate derived from the NCVS. To assume that even 24% of respondents are actively trying to create a dramatic story which is internally consistent enough to deceive the experienced team of interviewers Kleck used, much less a sufficient number to account for the vast difference in estimates between the Kleck survey and the NCVS, is in itself hard to believe. For the NCVS to miss a DGU requires only that the respondent not volunteer the information, but as Kleck observes "[t]here is no precedent in criminological research" for the "enormous level of intentional and sustained falsification" on the part of respondents which would be required in order to account for the 30-fold difference in estimated annual DGUs between Kleck's survey and the NCVS.

Indeed, Kleck argues, there is reason to believe that the estimate of 2.5 million annual DGUs is too low. There is the problem of self-censorship in reporting, since there is evidence that some respondents to the Kleck survey were wary of reporting such incidents to anonymous strangers, much less to government officials as in the NCVS. Several respondents expressed suspicion of the interviewer prior to answering "no" to the DGU question, perhaps because they view their actions as legally questionable. There may also be instances in which respondents regarded their own DGU incidents as too minor to report, or households in which DGU incidents of other household members were not known to the respondent, or where it was not considered appropriate to discuss the DGU incidents of other household members. The limitations of a telephone survey preclude sampling among the 5% of American households without telephones, most of which are poor and/or rural. As poor people are more likely to be victims of crime, and there is a higher rate of gun ownership in rural areas, this could also contribute to an underestimation of the number of DGUs.

Recommended Reading:

"Armed Resistance to Crime: The Prevalence and Nature of Self-Defense with a Gun", Kleck, Gary and Gertz, Marc
J. of Criminal Law and Criminology, v.86, n.1, pp.150-187 (1995)

Gun Laws and the Need for Self-Defense (Parts 1 and 2), hearings before The Subcommittee on Crime of the Committee on the Judiciary, U.S. House of Representatives, 104th Congress, 1st session, March 31, 1996
SuDoc# Y 4.J89/1:104/43/Pt.1, and 2nd session, April 5, 1995, SuDoc# Y 4.J89/1:104/43/Pt.2 [These hearings, called in response to the fledgling Republican Congress' efforts to repeal the 1994 Clinton/Feinstein gun ban (see 3.3 and Appendix I), featured testimony from a number of scholars cited elsewhere in this FAQ, including James Wright, Joyce Malcolm, David Bordua, Robert Cottrol, and Daniel Polsby, as well as law enforcement officers and crime victims.]
See testimony of David Bordua and Garen Wintemute.

Sourcebook of Criminal Justice Statistics 1994, Bureau of Justice Statistics, Office of Justice Programs, U.S. Dept. of Justice
SuDoc# J 29.9/6:994 (1994)