The AMA, Firearms, and Intellectual Dishonesty

Robert J. Woolley, M.D.

May 1999

Abstract: The author exposes the blatant bias and intellectual dishonesty in a 1998 AMA publication titled: Physician Firearms Safety Guide which advised physicians to counsel their patients against gun ownership.

Any discussion of guns and gun control is likely to arouse intense, polarized feelings in a nation which must live simultaneously with levels of gun violence unparalleled in Western democracies and with the right to gun ownership enshrined in its constitution. Any American paying attention to public discourse knows that gun policy is as divisive as abortion or the death penalty. But when such issues are discussed among professionals and academics, one expects a high level of even-handedness and intellectual honesty. Unfortunately, this has not always been the case. Political and emotional considerations sometimes overwhelm scientific integrity.

In November, 1998, the American Medical Association published, with considerable fanfare, the "Physician Firearm Safety Guide", written by Roger L. Brown and Larry S. Goldman of the AMA staff, with the AMA's name, logo, and slogan on the cover. One would expect to find only the most well-documented, non-controversial, unbiased advice coming from such a respected organization.

Unlike some other physicians who share my views opposing most forms of gun control, I am willing to grant that violence, in all its subtypes, can legitimately be viewed as a public health issue. I will further grant that gun violence is, in theory, at least as valid an issue to discuss with patients as, say, the use of seat belts.

However, this issue is much more complex and has a more ambiguous risk/benefit assessment than most such other issues which physicians discuss with patients. The cost/benefit analysis of the use of seat belts, for example, is simple; with rare exceptions, one need not do any customized assessment of whether a given patient is better off with or without habitual seat belt use. Unfortunately the clinical recommendations set forth by Brown and Goldman depend entirely on the assessment being similarly simple, obvious, universal, and one-sided. Far from being a fair-minded assessment of the evidence relevant to counseling patients on the issue of firearms in the home, it is a political tract, one-sided from beginning to end.

Evidence of Bias

Evidence of the authors' intention to be advocates of one side of the controversy, rather than neutral judges of the evidence, is literally to be found from the first page to the last. Inside the front of the booklet is acknowledgement of 15 academics to whom the manuscript was submitted for comments. A Medline search reveals that without exception, every one of these reviewers who has published one or more articles with a subject heading of firearms has taken an overtly pro-gun-control stance. This obviously biased selection reveals unambiguously that the authors did not wish their work to be scrutinized by those who might differ from their preconceived ideas. Intellectually honest writers would prefer comment from those with opposite viewpoints, the better to catch errors and be challenged to present the evidence fairly.

At the end of the booklet, the list of resources contains, in addition to governmental bodies, 16 private or university-based organizations with an explicit anti-gun agenda, and only one organization opposed to gun control (the National Rifle Association). Had the authors been interested in even a semblance of balance, they could have filled the blank half-page at the end of this section with several other gun rights organizations.

Further evidence of the authors one-sided intentions comes from mention of funding from the Joyce Foundation, which is well know for supporting anti-gun causes. The Foundation would not have been likely to support a project which had conclusions significantly different from those at which Brown and Goldman arrived; their conclusions were, quite apparently, predestined.

I shall pass over much of the monograph's discussion of epidemiology and firearm basics, except to note that a physician who has little training or personal experience with firearms cannot possibly give intelligent counsel to a patient on the matter. On page 23 we get to a "risk assessment". This supposedly "allows a rough measurement of risk", by which is meant whether a patient is at high risk for being injured by or causing injury with a gun in the home. Glaringly omitted is any similar instrument for measuring whether the patient is at risk for violence being visited upon him, which might be repelled by proper use of a firearm.

Defensive Value of Guns

The reason for this omission becomes apparent four pages later, when we get to the central deception of the brochure: "Several careful studies show that the risk of harm at home far outweighs the benefits afforded by self-protection. Studies purporting to show the benefits of home firearm ownership have greatly inflated the number of times such firearms are used in self-defense to protect the home's inhabitants".

It is true that several studies claim to show a high risk associated with gun ownership. Whether they are sufficiently "careful" studies is hotly debated, a point which Brown and Goldman conveniently fail to mention. Many studies also show the benefits of gun ownership, but the authors dismiss these studies without giving any particular references.

There is considerable debate about how often guns are actually used (whether fired or not) in defense of self, others, or property. Kleck and Gertz, in 1995, identified 13 surveys that attempted to describe the incidence of defensive use of guns, varying widely in quality, size, subjects, time interval covered, and methodology. When statistically adjusted to make the measurements more or less comparable, the results were between 764,000 and 3,600,000 defensive gun uses (DGUs) per year in the United States. To this list, Kleck and Gertz added their own new survey, which yielded a best estimate of 2,550,000 DGUs per year, though with wide confidence intervals. The outlier among these studies is the ongoing National Crime Victimization Survey (NCVS), a Census Bureau study conducted for the Department of Justice, which typically reveals something on the order of 70,000 DGU per year.

Although there are myriad technical issues which might cause over- or under-estimates of the true number of DGUs, the largest debate is whether survey respondents over-reported to the anonymous phone surveys, or under-reported to the non-anonymous NCVS. Put most starkly, if the Kleck and Gertz estimate is correct, then about 97% of respondents are forgetting or concealing a DGU to the NCVS surveyors; if the NCVS is correct, then a similar proportion of Kleck and Gertz's positive respondents were concocting or at least exaggerating their claimed defensive gun uses.

Subsequent to publication of the Kleck and Gertz survey, three more have been put forth. Cook and Ludwig conducted the National Survey on Private Ownership of Guns under the auspices of the US Department of Justice. [1997] Their raw numbers of DGUs were astonishing: 23,000,000. This was likely inflated by a few outliers reporting implausible numbers of DGUs each. Even after correcting for this phenomenon, though, they were still left with 1.5 million people annually having 4.7 million DGUs. This was apparently embarrassing to the long-time pro-gun-control authors, who then spent much of the report's space trying to downplay their own findings.

Next, researchers with the Centers for Disease Control and Prevention published the results of a 1994 random-dialing telephone survey. [Ikeda 1997] Based on the responses, they projected that there are 498,000 episodes per year in the US wherein a person hears an intruder, retrieves a firearm, sees an intruder, and believes that the intruder was repelled by the presence of the gun. This is within a factor of two of one of Kleck and Gertz 's results, since about 900,000 of their reported DGUs occurred within the victim's home. Ikeda's single specific type of DGU dwarfs the estimate of all types of DGUs from the NCVS.

Most recently, Hemenway and Azrael [1998] report two similar telephone surveys. In their 1994 study, extrapolation yielded an estimate of about 900,000 DGUs against humans per year. (Incidentally, I cannot help wondering why Cook, Ludwig, and Hemenway spent more public dollars conducting three additional surveys of similar methodology, when they claim that such surveys will always grossly exaggerate the numbers they are seeking. It would be interesting to see if their grant applications admitted in advance that the data collected would be worthless.)

In spite of the rhetorical attack that gun-control advocates have launched against Kleck and Gertz's survey data, it remains the case that such criticisms are based on speculation. Although critics allege that the great majority of DGU reports are false, they have not offered any empirical data to explain why they believe this is true of the approximately eighteen such surveys that have now been conducted.

The upshot of this discussion is that it is by no means a foregone conclusion that studies demonstrating a large number of DGUs are greatly inflated or exaggerated, as Brown and Goldman insist. The debates, largely centering on technical methodological issues, on the correct number of DGUs run to over 130 pages published in various academic journals (not to mention unpublished papers, professional conference proceedings, books, Internet debate, etc.) just since 1995. Yet Brown and Goldman apparently have no qualms about taking 27 words to report the Kleck and Gertz estimate (2.5 million), and about five times that many to dismiss it, accepting critics' arguments without question, but with no notice of any arguments that might support the Kleck and Gertz work. Their summary of the evidence is grossly distorted and represents intellectual dishonesty of a high order.

At a minimum they owe their readers an acknowledgement that they are rejecting a quantitative majority of the available evidence, a comment on their basis for doing so, and a concession that thoughtful and credentialed academics disagree with their judgment as to where the best evidence points.

Deterrent Effect Ignored

Even if we accept the bare minimum number of actual confrontational DGUs, Brown and Goldman completely ignore the general deterrent effect on crime that is caused by criminals' knowledge that roughly 40-50% of American households are armed. It is probably impossible to quantify this effect, but interviews with convicts clearly demonstrate that it is real. [Wright and Rossi, 1986, 1994] Even Cook and Ludwig, who generally share Brown and Goldman 's distaste for guns, comment: "Second and more generally, the number of DGUs tells us little about the most important effects on crime of widespread gun ownership. When a high percentage of homes, vehicles, and even purses contain guns, that presumably has an important effect on the behavior of predatory criminals. Some may be deterred or diverted to other types of crime...Such consequences presumably have an important effect on criminal victimization rates but are in no way reflected in the DGU count."

The point is that Brown and Goldman dismiss out of hand any significant beneficial effects of gun ownership. Once that is done, the risk/benefit analysis is conveniently reduced to a simple calculation similar to the seat belt case mentioned previously.

Which households are at risk?

Brown and Goldman devote one section of their tract to helping physicians identify patients or families who are at high risk for death or serious injury from firearms, but no comparable section to help identify patients who are at high risk for external threats for which a defensive firearm might prove beneficial. The authors are either deliberately withholding half of their analysis, or are so blinded by their prejudices that they fail to notice their own one-sidedness.

 Surely we can be more sophisticated than this. To any intellectually honest observer, it is obvious that there are households for which firearm ownership presents more risks than benefits-for example, a family in a low-crime neighborhood with a consistently rapid police response, with one or more family members with major depression and recent serious suicide attempts. On the other hand, for a family with no children or adolescents, no history of depression, violence, or suicide attempts, and adults trained and experienced in the use of their firearms, living in a high-crime area, it is highly probable that the benefits of gun ownership will outweigh the risks.

In their headlong rush to rid the world of all guns, Brown and Goldman simply ignore the existence of any such households. I do not claim to know what fraction of US households have more potential benefit than risk from owning a firearm, but it is clear that the "one size fits all" solution-turn in your weapons-is ill-suited for many households. Physicians who pass Brown and Goldman's simplistic advice to all their patients will be doing at least a sizeable portion of them a disservice.

There are no guns in this house.

Since Brown and Goldman are convinced that firearms in the home are a net liability to the family and have no defensive or deterrent utility, one would expect that these authors will have proudly displayed near their homes' front door a sign announcing that: "there are no guns in this house." After all, this would reduce the risk that criminals would break in for the purpose of stealing guns, without exposing the authors' families to any increased risk, since they apparently believe that criminals are not deterred by the presence of firearms (and hence would also not be encouraged by the absence of firearms). If they do not have such signs, one wonders if they are hypocritically depending on the general deterrent effect of the guns owned by their neighbors, since the predators do not know which houses do or do not have weapons.

Asking the Wrong Question

There are other areas where Brown and Goldman deceptively simplify complex and controversial issues. For example, "It is well-established that suicides and homicides occur more commonly in homes with firearms than in homes without them." This carefully worded statement conveys to the uninformed reader the sense that the presence of a firearm is a causal agent, while coming just short of saying so explicitly.

In another such slyly-worded claim, Brown and Goldman assert that "A firearm in the home is more likely to result in a death during a household quarrel, a suicide attempt, or an unauthorized shooting than in protecting members of the household. This may be true in the most limited sense, but even if so is of virtually no value. What is desired is to stop the felonious assault, not to kill the felon. To present the matter as Brown and Goldman do is to dismiss, by silence, the vast majority of cases in which a gun is used defensively with no shot being fired, a missed shot, or an injured but alive criminal. This fallacy has been debunked so often in print that it is inconceivable that Brown and Goldman were unaware of it. Their repeating it is therefore almost certainly another instance of deliberate shading of the truth.

Similarly, Brown and Goldman say: "Physicians can explain to patients that while many people feel safer with a gun in the home, the greatest risk of death from these firearms come [sic] from household members, intentional self-inflicted injuries, or unintentional discharge (especially by children)." Like the previously-examined claim, this one is probably true if one observes the careful parsing of words, but it provides the answer to the wrong question. The proper question is whether the possession of a firearm will increase or decrease the likelihood that a patient or a member of their family will be killed or injured.

To its credit, the Journal of the American Medical Association recently featured a pair of articles [Kleck - Aug. 98, Cummings and Koepsell - Aug. 98] which clearly explain how to pose the question in the correct way. To their discredit, Brown and Goldman do not, even though they cite the JAMA articles in their sources and therefore must have known of them. As the articles by Cummings and Koepsell, and by Kleck demonstrate, reasonable people can weigh the evidence in ways that point to opposite answers to this question (perhaps because a gun purchase raises the risk of death or injury for some people and lowers it for others). But one cannot arrive at a reasonable answer if one does not first pose the question in a reasonable way.

Inconsistent Evaluation of Evidence

In addition to selective presentation of evidence, Brown and Goldman are guilty of inconsistent evaluation of that evidence. For example, when discussing children's access to guns in the home, they say, "One cannot eliminate younger children's curiosity about guns, but one can reduce the likelihood that children will encounter them (there is little evidence that educating children to stay away from firearms is effective)." Similarly, "There is no evidence that such [firearm] training reduces the risk of injury or death from firearms." Notice that when a potential intervention (educating the family, but keeping the guns) is not to the authors' liking, the absence of evidence for that intervention is taken to argue against implementing it.

On the other hand, Brown and Goldman introduce a list of no fewer than 27 "public health interventions" related to guns with this statement: "Table 9 lists types of interventions that have been valuable in other public health problems and examples of how they might be applied to gun safety. Note that almost none of these interventions have been implemented widely or tested, but they certainly merit discussion because of successes with similar approaches to other problems." In other words, when a proposed intervention fits the authors' biased agenda, the absence of evidence is no hindrance at all.

This leads us to a consideration of the interventions Brown and Goldman propose. As with nearly everything else in their essay, the list reveals their bias. They propose that guns be made with a "lowered firing rate" and "lowered caliber ammunition", but without "recoil compensators" and "laser aiming devices". They want trigger locks, additional safety mechanisms, and less effective bullets.

Elsewhere in their tract (twice, actually) Brown and Goldman advocate keeping all guns in the house stored locked and unloaded. What all of these have in common is that they would make firearms less useful as defensive weapons. One can rationally argue for making guns owned by law-abiding citizens less effective for their intended task only if one has already concluded that their value for this purpose is negligible, as Brown and Goldman have clearly done, contrary to voluminous evidence.

Legislative Agenda

It should be noted, too, that this list implies an unspoken legislative agenda of mandating these items, since gun purchasers can already implement every item on Brown and Goldman's list of mechanical interventions if they choose to do so. That, of course, is unlikely as few gun owners would think it rational to deliberately reduce the effectiveness of their defensive weapons. I assume that Brown and Goldman would grant this point, since they have already concluded that anybody possessing a gun for defense is acting irrationally.


The AMA has lent its name, logo, prestige, and funding to the production of "educational" information for its members and their patients (the booklet includes a tear-out sheet to give to them) which is scientifically unsound, politically biased, and intellectually dishonest. It is an embarrassment to have a professional association which declares itself dedicated to science produce a publication that falls so short of basic academic tenets of honesty and fairness. The authors of the Physician Firearm Safety Guide clearly believe that their point of view would not be persuasive if they presented the evidence in an even-handed manner and allowed their readers to evaluate the evidence on their own. AMA members might consider whether they approve of their dues being spent to produce propaganda that so insults their intelligence.

Dr. Robert J. Woolley is a family practice physician at the University of Minnesota