UPDATE: The paper discussed in this post below has been withdrawn by the author Gary Kleck after Reason brought to his attention an important detail first pointed out by Robert VerBruggen of National Review: Kleck in the original paper treats the CDC's surveys on defensive gun use as if they were national in scope, as Kleck's original survey was, but they were not. From VerBruggen's own looks at CDC's raw data, it seems that over the course of the three years, the following 15 states were surveyed: Alaska, Colorado, Hawaii, Kentucky, Louisiana, Maryland, Mississippi, Montana, Ohio, New Hampshire, New Jersey, New York, North Dakota, Pennsylvania, and West Virginia. (Those states, from 2000 census data, contained around 27 percent of the U.S. population.) Kleck says he is working on a new version of the paper that recalculates the degree to which CDC's survey work indeed matches or corroborates his, and we will publish a discussion of those fresh results when they are complete. But for now Kleck has pulled the original paper from the web pending his rethinking the data and his conclusions.

Many people who support gun control are angry that the federal Centers for Disease Control (CDC) are not legally allowed to use money from Congress to do research whose purpose is "to advocate or promote gun control." (This is not the same as doing no research into gun violence, though it seems to discourage many potential recipients of CDC money.)

But in the 1990s, the CDC itself did look into one of the more controversial questions in gun social science: How often do innocent Americans use guns in self-defense, and how does that compare to the harms guns can cause in the hands of violent criminals?


Florida State University criminologist Gary Kleck conducted the most thorough previously known survey data on the question in the 1990s. His study, which has been harshly disputed in pro-gun-control quarters, indicated that there were more than 2.2 million such defensive uses of guns (DGUs) in America a year.

Now Kleck has unearthed some lost CDC survey data on the question. The CDC essentially confirmed Kleck's results. But Kleck didn't know about that until now, because the CDC never reported what it found.

Kleck's new paper—"What Do CDC's Surveys Say About the Frequency of Defensive Gun Uses?"**—finds that the agency had asked about DGUs in its Behavioral Risk Factor Surveillance System in 1996, 1997, and 1998.

Those polls, Kleck writes,

are high-quality telephone surveys of enormous probability samples of U.S. adults, asking about a wide range of health-related topics. Those that addressed DGU asked more people about this topic than any other surveys conducted before or since. For example, the 1996 survey asked the DGU question of 5,484 people. The next-largest number questioned about DGU was 4,977 by Kleck and Gertz (1995), and sample sizes were much smaller in all the rest of surveys on the topic (Kleck 2001).

Kleck was impressed with how well the survey worded its question: "During the last 12 months, have you confronted another person with a firearm, even if you did not fire it, to protect yourself, your property, or someone else?" Respondents were told to leave out incidents from occupations, like policing, where using firearms is part of the job. Kleck is impressed with how the question excludes animals but includes DGUs outside the home as well as within it.

Kleck is less impressed with the fact that the question was only asked of people who admitted to owning guns in their home earlier in the survey, and that they asked no follow-up questions regarding the specific nature of the DGU incident.

From Kleck's own surveys, he found that only 79 percent of those who reported a DGU "had also reported a gun in their household at the time of the interview," so he thinks whatever numbers the CDC found need to be revised upward to account for that. (Kleck speculates that CDC showed a sudden interest in the question of DGUs starting in 1996 because Kleck's own famous/notorious survey had been published in 1995.)

At any rate, Kleck downloaded the datasets for those three years and found that the "weighted percent who reported a DGU...was 1.3% in 1996, 0.9% in 1997, 1.0% in 1998, and 1.07% in all three surveys combined."

Kleck figures if you do the adjustment upward he thinks necessary for those who had DGU incidents without personally owning a gun in the home at the time of the survey, and then the adjustment downward he thinks necessary because CDC didn't do detailed follow-ups to confirm the nature of the incident, you get 1.24 percent, a close match to his own 1.326 percent figure.

He concludes that the small difference between his estimate and the CDC's "can be attributed to declining rates of violent crime, which accounts for most DGUs. With fewer occasions for self-defense in the form of violent victimizations, one would expect fewer DGUs."

Kleck further details how much these CDC surveys confirmed his own controversial work:

The final adjusted prevalence of 1.24% therefore implies that in an average year during 1996–1998, 2.46 million U.S. adults used a gun for self-defense. This estimate, based on an enormous sample of 12,870 cases (unweighted) in a nationally representative sample, strongly confirms the 2.5 million past-12-months estimate obtained Kleck and Gertz (1995)....CDC's results, then, imply that guns were used defensively by victims about 3.6 times as often as they were used offensively by criminals.

For those who wonder exactly how purely scientific CDC researchers are likely to be about issues of gun violence that implicate policy, Kleck notes that "CDC never reported the results of those surveys, does not report on their website any estimates of DGU frequency, and does not even acknowledge that they ever asked about the topic in any of their surveys."

NPR revisited the DGU controversy last week, with a thin piece that backs the National Crime Victimization Survey's lowball estimate of around 100,000 such uses a year. NPR seemed unaware of those CDC surveys.

For a more thorough take, see my 2015 article "How to Count the Defensive Use of Guns." That piece more thoroughly explains the likely reasons why the available DGU estimates differ so hugely.

However interesting attempts to estimate the inherently uncountable social phenomenon of innocent DGUs (while remembering that defensive gun use generally does not mean defensive gun firing, indeed it likely only means that less than a quarter of the time), when it comes to public policy, no individual's right to armed self-defense should be up for grabs merely because a social scientist isn't convinced a satisfyingly large enough number of other Americans have defended themselves with a gun.