Book Review

A few years ago I wrote a critical review of John Lott's book “More Guns Less Crime” (Hemenway 1998). Unfortunately, Lott's new book “The Bias Against Guns: Why Almost Everything You’ve Heard about Gun Control is Wrong” also has major problems. Much that Lott writes is either wrong or misleading. What follows is a partial listing:

1. Lott'states that: “the few existing studies that test for the impact of gun control laws on total suicide use purely cross-sectional level data, and find no significant relationship.” (p. 143). This statement is simply wrong. Lott must not have read the articles by Lester & Murrell 1982; Medoff & Magaddino 1983; Lester & Murrell 1986; Boor & Bair 1990; Yang & Lester 1991; Loftin et al 1991; or Carrington & Moyer 1994, all of which find a significant negative relationship between gun control laws and suicide rates. Lott also must have missed the various review articles on guns and suicide (e.g. Miller & Hemenway 1999; Brent 2001).

2. Lott'states that: “accidental shooters are overwhelmingly adult males with long histories of arrests for violent crimes, alcoholism, suspended or revoked drivers licenses, and involvement in automobile crashes.”(p. 83) This claim is completely misleading. The main empirical evidence Lott presents (p. 142) is from a 1973 article of 34 Vermont shooters in 1967 who caused unintentional firearm injuries to themselves or others (Waller & Worton 1973). The study actually found that 68% of these shooters had no evidence of violence, 79% had no evidence of alcohol problems, 81% had never had a drivers license suspended or revoked, and 38% had never even been involved as a driver in any type of traffic accident.

3. Lott'states that: “In the U.S. the states with the highest gun ownership rates also have by far the lowest violent crime rates.” (p. 76) He provides no evidence, no citations and no discussion for this assertion. The reader would never know that there is, in fact, an extensive literature examining the relationship between gun ownership rates and homicide (the most serious violent crime, the one with the most comparable and reliable data across states, and the crime where gun availability should have the most effect). Many of these studies find a significant positive relationship between gun ownership levels and homicide; where there are more guns there is more homicide (e.g., Brearly 1932; Seitz 1972; Lester 1990; Birckmayer 1999; Miller et al 20002; Hepburn & Hemenway 2003). I don’t believe that any peer-reviewed study finds a significant negative relationship between state gun levels and homicide.

4. Lott claims, with some justification, that some advocacy websites give false or misleading information. “In 2002, the National Crime Prevention Council claimed that ‘A child between ten and nineteen years old commits suicide with a handgun every six hours.’ During 1999, the Centers for Disease Control identified exactly two hundred handgun suicides for this age group, a rate of one every forty-four hours.” (p. 59). But Lott is misleading; unless you read his footnote, you don’t know that the CDC does not have information on the type of firearm used for most gun suicides. Lott does not tell the reader that in 1999 there were 1,078 firearm suicides, a rate of gun (but not handgun) suicide about every 8 hours. For 1994, CDC reported 1,564 firearms suicides among 10-19 year olds, a rate of firearm suicide every 6 hours.

5. Lott begins his chapter on guns in the media talking about concealed carry laws. He very briefly mentions his 2002 self-defense gun survey, and soon asks “why did the torrential news coverage of public school shootings in the 1990s fail to acknowledge when attacks were halted by citizens* with guns? A similar example of selective reporting occurred during January 2002 in a shooting that left three dead at the Appalachian Law School in Virginia.” Lott'spends three pages discussing this incident (pp. 24-27). He complains that the media did not pay enough attention to the fact that “the attack was stopped by two students who had guns in their cars.” Lott does say that “many stories mentioned the law enforcement or military background of these student heroes,” but not their guns. But Lott fails to tell the whole story and thus presents a misleading picture. It seems the shooting actually stopped because the shooter ran out of ammunition. And it appears that both students were actually police officers (http://www.cse.unsw.edu.au/~lambert/guns/roshhuntress.html). One not only retrieved a gun from his car but also a bulletproof vest and handcuffs.

6. Lott's book is largely about the benefits of self-defense gun use. Yet Lott misses most of the self-defense gun use literature. For example, he does not even mention the three national telephone surveys focusing on self-defense gun use sponsored by the Harvard Injury Control Research Center (Hemenway & Azrael 2000; Hemenway et al 2000). He discusses one of our thirty-plus empirical articles on gun issues, and he gets that one completely wrong.

We show that across regions and states, where there are higher levels of firearm ownership there are more suicides, more homicides, and more accidental (gun) deaths to children (and adults, and males and females), holding various factors constant, including poverty, urbanization, educational levels (alcohol consumption and non-lethal violent crime). (Miller et al, 2001; 2002a, 2002b,2002c,2002d).

Lott writes this about our study of 5-14 year olds: “The biggest problem is how the study measures what gun ownership rates are. The first two measures used were: 1) the adult firearm homicide and firearm suicide rates and 2) the adult firearm suicide rate, under the assumption that those rates are higher where guns are more common. Unfortunately, juvenile firearm homicide or suicides could be related to those measures for reasons unrelated to gun ownership. Assume two areas have the same gun ownership rates, if one had more adult firearm homicides, is it really surprising that it would also have more juvenile firearm homicides?” (p. 316).

While Lott's logic is correct, he must not have taken the time to actually read our study. We did NOT use either of these measures for gun ownership rates. We do use a variety of measures; including one that we carefully validate as the best measure (Azrael et al 2001). It is disheartening to do careful work and have Lott completely misrepresent it.

Lott proceeds to use the National Opinion Research Center (NORC) survey data to measure household gun ownership rates, apparently for each state for each year (p. 255). Yet the NORC survey cannot legitimately be used in this manner. The survey is not designed to give a representative sample for even one of the 50 states. For example, in North Dakota, all the survey respondents come from one county, and it is the same county in virtually every survey. More important, even if the survey were representative for states, it is far too small to give a reasonable estimate each year for the least populous states. Until recently, there were only about 1,500 respondents, or 30 per state; in states like North Dakota that amounted to perhaps 10 people (conversation with Tom Smith, NORC, 3/20/03).

Lott not only uses questionable measures of gun ownership but then he inappropriately adds 44 explanatory variables to his state-level analysis, an analysis that is largely cross-sectional (n=50) (it is largely cross-sectional since gun ownership rates change very slowly over time, and the best measures of gun ownership probably have confidence intervals wider than the yearly changes). Lott's approach virtually ensures that, whatever the true relationship between guns and death, his analysis will not find them.

7. Lott gives econometrics (sophisticated statistical analysis) a bad name. Most readers of The Bias Against Guns will not be trained statisticians. They will not be able to understand that Lott's methodology is often highly questionable. But readers will hopefully be able to see how absurd his results are. I believe most criminologists would expect that state laws on gun carrying would have no more than a very small impact on crime rates. Yet Lott finds enormous impacts. His analysis finds that permissive carry laws reduce murders from rampage killings by 75%! (p. 113). A more careful econometric analysis finds no effect of the gun carrying laws on rampage killings (Duwe et al 2002).

8. I believe that most criminologists would expect that state laws on gun shows and assault weapons would each have no more than a very small impact on crime rates. Yet Lott finds enormous impacts. Lott's analyses indicate that the “impact from closing the gun show ‘loophole’” was a reduction of 72% in Indiana’s violent crime rate and a reduction of 102% (!) in Indiana’s auto theft rate, while the effect on New York was to increase the violent crime rate by 83% and the auto theft rare by 34%. According to Lott's model, the impact of banning assault weapons in Hawaii was that its violent crime rate increased 55%, and its robbery rate rose 95%! By contrast, the assault weapons ban reduced Maryland auto theft rate by 57% (Table 8.7; pp. 216-17). Of course, none of these incredible things happened [e.g. Indiana did not experience what would have been a crime miracle—(e.g., a negative car theft rate, which must mean thieves returning cars)—following the closing of the gun show loophole, nor did New York experience a crime spree of almost unprecedented proportions). Lott's results are just one piece of evidence that his models are misspecified, and should not be accepted as valid.

9. Lott also tries to determine the effects of state safe storage laws. The earliest date a state law went into effect was in October 1989 in Florida. Lott'says he uses a poll from 1994 that asked about gun storage to determine the effect of the laws. His results “indicate that states with safe storage laws had higher rates at which households left guns loaded and unlocked but that the rate fell the longer that the law was in effect. Six years* after adoption of the law, states with safe storage laws have a lower percentage of homes with loaded, locked guns than do states without those laws.” (p. 175) Yet in 1994, when the data were collected, only one state had experienced the law for 5 years, and none for six years.

10. Lott has been under attack concerning a mysterious 1997 survey related to his repeated claims that in 98% of defensive gun uses the defender merely has to brandish the gun to break off an attack. Nine published surveys give numbers ranging from 21% to 67% as to how often defenders shoot, far more than the 2% Lott claims. Lott repeatedly changed his story about the source of the 98% figure. He now claims he conducted a survey over three months in 1997, but he has not been able to present any credible evidence that he conducted the survey. And Lott made the 98% claim on February 6, 1997, well before his purported survey could have been complete (http://www.cse.unsw.edu.au/~lambert/guns/lott98update.html).

In The Bias Against Guns Lott reports on a survey he conducted in November/December 2002, supposedly to validate his missing survey. He writes: “…the survey I conducted during the fall of 2002 indicates that simply brandishing a gun stops crime 95% of the time, and other surveys have also found high rates” (p. 24). He describes his survey on the last pages of the book. Exactly 1,015 people were interviewed (p. 259). But unfortunately Lott's book abruptly ends before he gives the survey results. Internet reports say that 7 of his 1,015 respondents claimed a self-defense gun use (13 uses), and in one of these the respondent shot his gun. This is not 98% or 95%. Most important, his survey is not nearly large enough to provide precise estimates of the percentage of self-defense gun users who merely brandish the firearm.

In his analyses, Lott virtually always uses complicated econometrics. For readers to accept the results requires complete faith in Lott's integrity, that he will always conduct careful and competent research. Lott does not merit such faith. It is unfortunate, for Lott analyzes important policy issues in a contentious policy arena where more good research is needed. It is also disheartening for the many careful academic researchers who are trying to conduct competent studies to see the impact that Lott's less careful research has had on the policy debate.

*All underlines have been added

References:

Azrael D, Cook PJ, Miller M. State and local prevalence of firearms ownership: measurement, structure and trends. National Bureau of Economic Research Working Paper, #8570, 2001.

Birckmayer J. The Role of alcohol and Firearms in Youth suicide and Homicide in the United States. PhD dissertation. Harvard School of Public Health. 1999.

Boor M, Bair JH. Suicide rates, handgun control laws, and sociodemographic variables. Psychological Reports. 1990; 66:923-30.

Brearly HC. Homicide in the United States. Chapel Hill: University of North Carolina Press, 1932.

Brent DA. Firearms and suicide. Annals of the New York Academy of Sciences. 2001; 932:225-39.

Carrington PJ, Moyer S. Gun control and suicide in Ontario. American Journal of Psychiatry. 1994; 151:606-608.

Duwe G, Kovandzic T, Moody CE. The impact of right-to-carry concealed firearm laws on mass public shootings. Homicide Studies Journal. 2002; 6:271-296.

Hemenway D. Book Reviews. Lott JR, Jr. More Guns Less Crime; Diaz T. Making a Killing. New England Journal of Medicine. 1998; 339:2029-30.

Hemenway D, Azrael DR. The relative frequency of offensive and defensive gun use: Results from a national survey. Violence and Victims. 2000; 15:257-272.

Hemenway D, Miller M, Azrael DR. Gun use in the United States: Results from two national surveys. Injury Prevention. 2000; 6:263-267.

Hepburn LM, Hemenway D. Firearm availability and homicide: A review of the literature. Harvard Injury Control Research Center 2003.

Lester D. Relationship between firearm availability and primary and secondary murder. Psychological Reports. 1990; 67:490.

Lester D, Murrell ME. The preventive effect of strict gun control laws on suicide and homicide. Suicide and Life Threatening Behavior. 1982; 12:131-140.

Lester D, Murrell ME. The influence of gun control laws on personal violence. Journal of Community Psychology. 1986; 14:315-18.

Loftin C, McDowall D, Wiersema B, Talbert CJ. Effects of restrictive licensing of handguns on homicide and suicide in the District of Columbia. New England Journal of Medicine. 1991; 325:1615-1620.

Medoff MH, Magaddino JP. Suicide and firearm control laws. Evaluation Review. 1983; 75:357-372.

Miller M, Azrael D, Hemenway D. Firearm availability and unintentional firearm deaths. Accident Analysis and Prevention. 2001; 33:477-484.

Miller M, Azrael D, Hemenway D. Firearm availability and unintentional firearm deaths, suicide and homicide among 5-14 year olds. Journal of Trauma. 2002b; 52:267-75.

Miller M, Azreal D, Hemenway D. Firearm availability and unintentional firearm deaths, suicide and homicide among women. Journal of Urban Health. 2002c; 79:26-38.

Miller M, Azrael D, Hemenway D. Household firearm ownership levels and homicide across U.S. regions and states, 1988-1997. American Journal of Public Health. 2002a; 92:1988-1993.

Miller M, Azrael D, Hemenway D. Household firearm ownership levels and suicide across U.S. regions and states, 1988-1997. Epidemiology. 2002d; 13:517-524.

Miller M, Hemenway D. The relationship between firearms and suicide; A review of the literature. Aggression and Violent Behavior. 1999; 4:59-75.

Seitz ST. Firearms, homicide and gun control effectiveness. Law and Society Review. 1972; 6:595-614.

Waller JA, Whorton EB. Unintentional shootings, highway crashes and acts of violence. Accident Analysis and Prevention. 1973; 5:351-56.

Yang B, Lester D. The effect of gun availability on suicide rates. Atlantic Economic Review. 1991; 19:74

HARVARD SCHOOL OF PUBLIC HEALTH

Page generated: Friday, August 8, 2003

Copyright, 2002, President and Fellows of Harvard College