From: [REDACTED] at [hprnd.rose.hp.com] (Steve Kao) Newsgroups: talk.politics.guns Subject: Re: anti-gun myths? Faq somewhere? Date: 28 Oct 1993 00:23:04 GMT The following was compiled by many contributors to talk.politics.guns. - Steve Kao ----------------------------------------------------------------------- "Protection or Peril? An Analysis of Firearm-Related Deaths in the Home," Arthur L. Kellermann and Donald T. Reay, The New England Journal of Medicine 314, no., 24 (June 12, 1986): 1557-1560. The article is reprinted in: The Gun Control Debate, You Decide ed. Lee Nisbet, Prometheus Books, 1990, 239-244. Procedure: The medical examiner case files for every firearm related death in King County, Washington (1980 population = 1,270,000 including Seattle = 494,000 and Bellevue = 74,000) between January 1, 1978 and December 31, 1983 was reviewed. Incomplete records were corroborated with information from police case files and interviews of investigating officers. "Gunshot deaths involving the intentional shooting of one person by another were considered homocides. Self-protection homocides were considered "justifiable" if they involved the killing of a felon during the commission of a crime; they were considered "self-defense" if that was the determination of the investigating police department and the King County prosecutor's office. All homocides resulting in criminal charges and all unsolved homocides were considered criminal homocides." Data: 6 year period 743 deaths from firearms (= 9.75 / 100,000 per year) = 22.7 percent of all violent deaths in King County excluding traffic deaths = 45% of all homocides (national avg. = 61%) = 49% of all suicides (national avg. = 57%) = <1% of all accidental deaths = 5.7% of deaths in undetermined circumstances inside a house or dwelling = 473 deaths (63.7%) in the home where the firearm involved was kept = 398 (53.6%) breakdown of 398 deaths in home where gun was kept: suicides = 333 (83.7%) homocides = 50 (12.6%) accidents = 12 (3%) unknown = 3 (0.7) breakdown of suicides with guns in home where gun was kept: male victim = 265 (80%) female victim = 68 (20%) blood tested for ethanol = 245 (74%) blood alcohol test positive = 86 (35% of those tested) blood alcohol level above 100 mg/dl = 60 (24.5% of those tested) handgun used = 226 (68%) long gun used = 107 (32%) breakdown of homocides with guns in home where gun was kept: male victim = 30 (60%) female victim = 20 (40%) blood tested for ethanol = 47 (94%) blood alcohol test positive = 27 (54% of those tested) blood alcohol level above 100 mg/dl = 10 (21% of those tested) handgun used = 34 (68%) long gun used = 16 (32%) occurred during altercation in the home = 42 (84%) self-defense during altercation = 7 of 42 (17%) justifiable homomcide of burglars = 2 of 50 (4%) resulted in criminal charges = 41 of 50 (82%) total self-defense and justifiable = 9 of 50 (18%) breakdown of accidental deaths with guns in home where gun was kept: male victim = 12 (100%) blood alcohol test positive = 2 (17%) handgun used = 11 (92%) deaths excluding suicides = 65 (50 homocide, 12 accident, 3 unknown) victim was stranger = 2 (3%) victim was friend or acquaintance = 24 (37%) victim was resident = 36 (55%) victim of homocide was resident = 29 (45% of total, 58% of homocides) resident shot by family member except spouse = 11 (31%) by spouse = 9 (25%) by self = 7 (19%) by roommate = 6 (17%) by other = 3 (8%) Conclusions: ratio of killed by stranger to killed by person known = 12:1 ratio of accidental deaths to self-protection homocides = 1.3:1 ratio of criminal homocides to self-protection homocides = 4.6:1 ratio of suicides to self-protection homocides = 37:1 ratio of suicides, criminal homocides, and accidental deaths to homocides for self-protection = 43:1 ------------------ end ------------ (Thanks to Robert I. Kesten) ============================================ Problems with this paper: 1) Acquaintances are listed as homicides, while the category includes abusive husbands, neighborhood gangs, and rival drug dealers. 2) There is no attempt to control for risk behavior among gun-owners. Households with residents with criminal or negligent histories are grouped with households lacking such histories, and the latter are smeared by the formers misdeeds. 3) The vast majority of deaths are suicides, and no attempt is made to account for possible substitution of other methods. 4) The most blatant problem with the statistic is that the 43:1 ratio is given as badness:goodness, while ignoring the fact that using a handgun to prevent a crime is NOT equivalent to killing an intruder. The vast majority of successful defenses with weapons do not involve firing (the threat alone is sufficient), and even if the gun is fired AND the intruder is hit, handgun wounds are only about 1/4 lethal. One can divide the one million defensive uses of guns/year by the 31,000 gun deaths/year (due to all causes) and conclude that a gun is 33 times as likely to be used in thwarting a crime than to kill someone. ============================================ The response is "You are quoting a 10-year old study done in only one county in the US? Just one?!? And your saying that's the rule nationwide in the USA and Canada!?!?! You are using a sample size of one, which any statistician or scientist will tell you, is rather meaningless. Besides that study ignores the fact that suicides (37 out of the 43) aren't effected by gun control, they just change methods as seen in Canada following C51 in 1978. It ignores the fact that for each single defensive homicide, there were roughly 9 defensive uses where the attacker is only wounded, and the 490 cases where the attacker is driven off unhurt (these are from a national US survey)." - Greg Booth ============================================ The largest flaw in this statistic is a basic misunderstanding of self-defence and its goals: The aim of self-defence is to prevent a crime, not to kill the attacker. In fact, under 0.5% of sucessful self-defence uses result in the death of the attacker (0.5% is based on ~400 justifiable homocides by civilians each year, FBI Uniform Crime Report, and the National Crime Survey's 80,000 self-defence uses which is a reasonable _minimum_ estimate of the number of civilian with-gun self-defences.) This factor of 200 error alone reverses the implications of the statistic, however this is not the only problem. The study looked exclusively at burglaries, and did not consider other crimes (attempted murder, rape, etc...). This probably represents another factor of five or ten. Then there is the question of self-defence against family members: If an abused wife were to shot her husband in self-defence, the study classified this as a "kill[ing] a friend or family member" (bad). In my opinion, preventing an attack, even by a (ex) friend or familty member is a good thing, not a bad one. If I had to guess (correcting the 43:1 figure, based on the above, etc...) a more accurate and informative statement would be: "A gun in the home is roughly ten times more likely to be used to prevent a crime, than harm a friend or family member." Even this, however, is not the complete picture: To be truely meaningfull, the statistic should compare the "average" person to the data in the study. As an absurd example, consider a person's risk of developing prostate cancer, assuming one lives long enough. If you were to simply look at the total number of cases, you would say a person has about a 40% risk of this. However, such a statistic would be highly misleading: About 80% of men, and 0% of women develop this cancer. To really learn anything from the numbers, you have to look more carefully. In the case of the 43:1 statistic, it is useful to know that in ~85% of the cases where someone is killed by a friend or a family member, there was a police record of violence (criminal records, police calls over "domestic disturbances, etc...) In other words, if you have no history of violence, you are about five times _less_ likely to kill a friend or family member. So, to restate the 43:1 number in its more meaningfull form: "If you or someone living with you has a history of violent behavior, a gun is (very roughly) equally likely to be used to prevent a crime or hurt a family member or friend. But, if you and the people you live with have no history of violence, a gun is (again very roughly) fifty times more likely to be used to prevent a crime, than to hurt a friend or family member." As for discussing this in a letter to the editor, I would ignore almost all of the above: Such a letter should be short, direct and focused on a single point. (This increases the chances of it being published, and decreases the risk of editing...) I would suggest pointing out the factor of 200 difference between killing attackers, which the statistic measures, and preventing crimes, which is what self-defence is really all about. If possible, within the paper's guidelines (usually, they say something to the effect that letters under n words are more likely to be printed), it might also be a good idea to mention the study's exclusive focus on burglaries (since most of us also care about defence against murderers, rapists, etc...) _or_ the disparity between people with a history of violence (who probably should avoid _all_ weapons, kitchen knives inculded) and those with no history of violence (who aren't a significant threat, however armed, to friends and family.) Frank Crary CU Boulder ============================================ Its a ten year old study done in 1 (only 1) county in the USA. Its not over several counties, or several states, just 1 county. Does that make its results valid to apply to other areas, including other countries? NO! - Stirling Chow ============================================ Here is a section quotes from a little pamphlet distributed at the local ranges. The pamphlet is titled "GUNS: Facts & Fallacies" and is authored under "Doctors for Integrity in Research & Public Policy", Edgar A. Suter, MD, Chairman. Phone # (510) 277 0333. "We have all heard that ``a gunowner is 43 times more likely to kill a family member than intruder.'' How did this fallacy start? In a 1986 article in the _New England Journal of Medicine_, Drs. Kellerman and Reay described the proper way to calculate how many people are saved by guns campared to how many are hurt by guns. The benefits should include, in the authors' own words ``cases in which burglars or intruders are wounded or frightened away by the use or display of a firearm [and] cases in which would-be intruders may have purposely avoided a house known to be armed...'' However, when Kellerman and Reay calculated their comparison, they didn't include those cases, they only counted the the times a homeowner (italics) killed the criminal. Because well under 1% of defensive gun usage involves the death of a criminal, _Kellerman_and_Reay_understated_ _the_protective_benefits_of_firearms_by_a_factor_of_at_least_100! They turned the truth on its head! Objective analysis, even by Kellerman and Reay's own standards, show the ``43 times'' comparison to be superficially appealing, but actually an empty contrivance - unfortunately one that is repearted by the well-funded gun-prohibitionist lobby and by biases journalists." (all emphasis as per the original) hope folks will find some use for this. dsa ============================================ Gun Stats & Mortal Risks Preston K. Covey Erik Larson's even-handed article on Paxton Quigley (_Armed_Force_, 2/4/93, WSJ) cites the world~s most notorious "statistic" regarding guns in the home: "A pioneering study of residential gunshot deaths in King County, Washington, found that a gun in the home was 43 times more likely to be used to kill its owner, spouse, a friend or child than to kill an intruder." The "43 times" stat is everywhere these days; it has grown in media lore like the proverbial urban myth: it was inflated by one pugilistic talk-show pundit to "93." Given the shock value of the finding, the conclusion of the 1986 New England Journal of Medicine (NEJM) study is remarkably understated: "The advisability of keeping firearms in the home for protection must be questioned." Responsible people should indeed question the risks and benefits of bringing a firearm into their home. But what we need to know is this: What exactly are the risks and benefits? The NEJM testimony is neither the whole truth about the benefits nor nothing but the truth about the risks. Further, as with motor vehicles, we want to know: What control do we have over the risks and benefits? And, as with the risks of cancer or heart disease or auto accidents: How can we minimize the risks? Like raw highway death tolls, the NEJM stat is not very helpful here. The NEJM finding purports to inform us, but it is framed to warn us off. It is widely promulgated in the media as a "scare stat," a misleading half-truth whose very formulation is calculated to prejudice and terrify. The frightful statistic screams for itself: The risks far outweigh the benefits, yes? What fool would run these risks? If your car were 43 times more likely to kill you, a loved one, a dear friend or an innocent child than to get you to your destination, should you not take the bus? Uncritical citation puts the good name of statistics in the bad company of lies and damned lies. Surely, we can do better where lives are at stake. Let's take a closer look at this risky business: The "43 times" stat of the NEJM study is the product of dividing the number of home intruders/aggressors justifiably killed in self-defense (the divisor) into the number of family members or acquaintances killed by a gun in the home (the dividend). The divisor of this risk equation is 9: in the study's five-year sample there were 2 intruders and 7 other cases of self-defense. The dividend is 387: in the study there were 12 accidental deaths, 42 criminal homicides, and 333 suicides. 387 divided by 9 yields 43. There were a total of 743 gun-related deaths in King County between 1978 and 1983, so the study leaves 347 deaths outside of homes unaccounted. The NEJM's notorious "43 times" statistic is seriously misleading on six counts: 1. The dividend is misleadingly characterized in the media: the "or acquaintances" of the study (who include your friendly drug dealers and neighborhood gang members) is equated to "friends." The implication is that the offending guns target and kill only beloved family members, dear friends, and innocent children. Deaths may all be equally tragic, but the character and circumstance of both victims and killers are relevant to the risk. These crucial risk factors are masked by the calculated impression that the death toll is generated by witless Waltons shooting dear friends and friendly neighbors. This is criminological hogwash. 2. The study itself does not distinguish households or environs populated by people with violent, criminal, or substance-abuse histories -- where the risk of death is very high -- versus households inhabited by more civil folk (for example, people who avoid high-risk activities like drug dealing, gang banging and wife beating) -- where the risk is very low indeed. In actuality, negligent adults allow fatal but avoidable accidents; and homicides are perpetrated mostly by people with histories of violence or abuse, people who are identifiably and certifiably at "high risk" for misadventure. To ignore these obvious risk factors in firearm accidents and homicides is as misleading as ignoring the role of alcohol in vehicular deaths: by tautology, neither gun deaths nor vehicular deaths would occur without firearms or vehicles; but the person and circumstance of the gun owner or driver crucially affect the risk. 3. One misleading implication of the way the NEJM stat is framed is that the mere presence of a gun in the home is much more likely to kill than to protect, and this obscures -- indeed, disregards -- the role of personal responsibility. The typical quotation of this study (unlike Larson's) attributes fatal agency to the gun: "A gun in the home is 43 times as likely to kill . . . ." (The Center to Prevent Handgun Violence, a major promulgator of the NEJM statistic, uses this particular formulation.) We can dispense with the silly debate about whether it's people or guns that accomplish the killing: again, by tautology, gun deaths would not occur without the guns. The question begged is how many deaths would occur anyway, without the guns. In any case, people are the death-dealing agents, the guns are their lethal instruments. The moral core of the personal risk factors in gun deaths are personal responsibility and choice. Due care and responsibility obviate gun accidents; human choice mediates homicide and suicide (by gun or otherwise). The choice to own a gun need not condemn a person to NEJM's high-risk pool. The gun does not create this risk by itself. People have a lot to say about what risk they run with guns in their homes. For example, graduates of Paxton Quigley's personal protection course do not run the touted "43 times" risk any more than skilled and sober drivers run the same risks of causing or suffering vehicular death as do reckless or drunk drivers. Undiscriminating actuarials disregard and obscure the role of personal responsibility and choice, just as they disregard and obscure the role of socio-economic, criminological and other risk-relevant factors in firearm-related death. This is why we resent insurance premiums and actuarial consigment to risk pools whose norms disregard our individualities. Fortunately, nothing can consign us to the NEJM risk pool but our own lack of choice or responsibility in the matter. 4. Suicide accounts for 84% of the deaths by gun in the home in the NEJM study. As against the total deaths by gun in King County, including those outside the home, in-house suicides are 44% of the total death toll, which is closer to the roughly 50% proportion found by other studies. Suicide is a social problem of a very different order from homicide or accidents. The implication of the NEJM study is that these suicides might not occur without readily available guns. It is true that attempted suicide by gun is likely to succeed. It is not obviously true that the absence of a gun would prevent any or all of these suicides. This is widely assumed or alleged, but the preponderance of research on guns and suicide actually shows otherwise, that this is wishful thinking in all but a few truly impulsive cases. (See: Bruce L. Danto et al., The Human Side of Homicide, Columbia University Press, 1982; Charles Rich et al., "Guns and Suicide," American Journal of Psychiatry, March 1990.) If suicides were removed from the dividend of the NEJM study's risk equation, the "43 times" stat would deflate to "six." The inclusion of suicides in the NEJM risk equation -- like the causes, durability, or interdiction of suicidal intent itself -- is a profoundly debatable matter. Quotations of the NEJM study totally disregard this issue. 5. Citations of the NEJM study also mislead regarding the estimable rate of justifiable and excusable homicide. Most measures, like the NEJM homicide rate, are based on the immediate disposition of cases. But many homicides initially ruled criminal are appealed and later ruled self-defense. In the literature on battered women, immediate case dispositions are notorious for under-representing the rate of justifiable or excusable homicide. Time's January 18, 1993, cover story on women "Fighting Back" reported one study's finding that 40% of women who appeal have their murder convictions thrown out. Time's July 17, 1989, cover story on a week of gun deaths reported 51% of the domestic cases as shootings by abuse victims; but only 3% of the homicides were reported as self-defense. In a May 14, 1990, update, Time reported that 12% of the homicides had eventually been ruled self-defense. In Time's sample, the originally reported rate of self-defense was in error by a factor of four. The possibility of such error is not acknowledged by promulgators of the NEJM statistic. 6. While both the dividend and the product of the NEJM risk equation are arguably inflated, the divisor is unconscionably misleading. The divisor of this equation counts only aggressors who are killed, not aggressors who are successfully thwarted without being killed or even shot at. The utility of armed self-defense is the other side of the coin from the harms done with guns in homes. What kind of moral idiocy is it to measure this utility only in terms of killings ? Do we measure the utility of our police solely in terms of felons killed -- as opposed to the many many more who are otherwise foiled, apprehended, or deterred? Should we not celebrate (let alone count ) those cases where no human life is lost as successful armed defenses? The question posed to media that cite the NEJM scare stat is this: Why neglect the compendious research on successful armed defense, notably by criminologist Gary Kleck (Point Blank: Guns and Violence in America , Aldine de Gruyter, 1992)? Kleck~s estimations of the rate and risk of defensive firearm use are based on victimization surveys as well as other studies: the rate is high (about one million a year) and the risk is good (gun defenders fare better than anyone, either those who resort to other forms of resistance or those who do not resist). Dividing one million gun defenses a year by 30,000 annual gun deaths (from self-defense, homicides, suicides, and accidents) yields 33. Thus, we can construct a much more favorable statistic than the NEJM scare stat: A gun is 33 times more likely to be used to defend against assault or other crime than to kill anybody. Of course, Kleck's critics belittle the dividend of this calculation; what is good news for gun defenders is bad news for gun control. We should indeed question the basis and method of Kleck's high estimation of defensive firearm use, as I have questioned the NEJM statistic. Clearly, the issue of how to manage mortal risks is not settled by uncritical citation of statistics. One thing troubles me still: we can hardly escape the unquestioned NEJM scare stat in our media, but we hardly ever find Kleck's good work mentioned, even critically. * * * ============================================ Then they quoted it wrong, three ways. The actual "statistic" is "a gun in the home is 43 times more likely to kill yourself or an acquaintance than an intruder." Problems with the statistic: 1) 37 of those 43 are suicides. Unless you think that buying a gun would make suicide start to look more attractive to you, to include these is just cooking the books. 2) "Dead burglars" don't measure the usefulness of a gun -- foiled crimes do. 98% of crimes foiled by a defensive gun are foiled with NO shots fired. Even when a shot is fired, a "dead burglar" results in only a small number of those cases. Cops aren't graded on their "dead burglar count" -- why should citizens be? 3) "Yourself or acquaintance" resolves to "anybody you know." Drug dealers tend to know other drug dealers; ex-wives tend to know their abusive ex-husbands, etc. 4) The 43 number also includes a significant number of legal self-defense homicides (e.g., the ex-wife killing the attacking ex-husband). 5) The numbers given are for all households with guns EVEN IF THE HOME GUN WAS NOT THE ONE INVOLVED IN THE SHOOTING. In other words, if an intruder catches you in the bathroom and shoots you while your gun sits quietly in the safe downstairs, you're counted as a victim of your nasty gun-in-the-home. ============================================ The original statistic was: "A gun in the home is more likely to kill you (i.e., including suicide) or someone you know (e.g., including an abusive husband or a rival drug dealer) than to kill (notice this -- KILL) an intruder." Subtracting suicides and justifiable homicides, and realizing that a gun is useful not only when it KILLS an intruder but when it STOPS A CRIME (even if never fired), some of the MEANINGFUL comparisons read as follows: A gun is 50 times more likely to be used to defend against criminal threat than to kill another person. A gun is 50 times more likely to be used to defend against criminal threat than to be used in suicide. A gun is 245 times more likely to be used by a non-criminal to defend against criminal threat than to commit criminal homicide. A gun is 535 times more likely to be used to defend against criminal threat than to accidentally kill anybody. -- [c d t] at [rocket.sw.stratus.com] --If you believe that I speak for my company, OR [c d t] at [vos.stratus.com] write today for my special Investors' Packet... ============================================