Statistical Malpractice: The Standard in Anti-Gun “Research”—Part 2

(from: youtube.com)

(from: youtube.com)

In Part I of this article, . . . we made some preliminary observations regarding the Harvard School of Public Health study published in the February 2002 issue of the Journal of Trauma. (1)

The Violence Policy Center (VPC) has been lauding the study as “the most comprehensive study ever conducted on impact of gun availability.” In its press release, the organization further states, “The elevated rate of violent death among children in high gun ownership states cannot be explained by differences in state levels of poverty, education, or urbanization.” (2)

The authors of the study assert in their abstract, “A statistically significant association exists between gun availability and the rates of unintentional firearm deaths, homicides, and suicides. The elevated rates of suicide and homicide among children living in states with more guns is not entirely explained by a state’s poverty, education, or urbanization and is driven by lethal firearm violence, not by lethal nonfirearm violence.” (1)

Now we will be examining the factors that Miller et al. claim were “not entirely” responsible for the high rates of unintentional firearm injury, homicide, suicide and overall violence in the mostly Southern states. Rather than using the biased VPC shibboleths “highest” or “lowest gun ownership states,” I have used the more objective terminology “high” and “low levels of juvenile violence” states for the purpose of this critique.

Although it’s not politically correct to bring race and ethnicity into the debate, let us state that illiteracy, broken homes and violence, with or without firearms, has been shown to be more prevalent, whether linked to cultural, socioeconomic or other factors, to African-Americans and Hispanics (and less so to whites and Asians). (3)

A Picture Worth a Thousand Words

The percentage of African-American (and other significant minority) students enrolled in public schools in those states with high levels of juvenile violence are as follows:

Louisiana, 47.6 percent (Hispanics, 1.3 percent)

Alabama, 36.4 percent (Hispanics, 1.1 percent)

Mississippi, 51 percent

Arkansas, 23.5 percent (Hispanics, 3.0 percent)

West Virginia, 4.2 percent.

The percentage of other minorities in these states can be assumed to be insignificant because they are so small (i.e., less than 1 percent).

On the other hand, the percentage of African-American (and other significant minority) students in states with low levels of juvenile violence are as follows:

Hawaii, 2.4 percent (Hispanics, 4.6 percent; Asians, 72.2 percent)

Rhode Island, 7.7 percent (Hispanics, 13.1 percent; Asians, 3.2 percent)

New Jersey, 18.1 percent (Hispanics, 14.9 percent; Asians, 6.1 percent)

Delaware, 30.6 percent (Hispanics, 5.4 percent; Asians, 2.2 percent)

Massachusetts, 8.6 percent (Hispanics, 10.2 percent; Asians, 4.3 percent). (4)

As can be seen by these figures, New Jersey does have a relatively high percentage of Hispanics and a small Asian minority. Hawaii has only 2.4 percent blacks, while whites are 20.5 percent, and Asians, the least violent ethnic group, constitute 72.2 percent of the population.

Politically incorrect as it may be, blacks and Hispanics account for a much higher percentage of students in high levels of juvenile violence states, at 33.6 percent, whereas whites and Asians account for 66.4 percent.

Conversely, in the low levels of juvenile violence states, blacks and Hispanics account for only 23.1 percent, whereas whites and Asians account for a larger percentage of the enrolled students in public schools at 76.9 percent.

No matter their race though, they deserve to be able to defend themselves in court so if your child has allegedly committed an act of juvenile violence then you may need to contact someone like these philadelphia criminal lawyers. Hopefully, then a rightful verdict will be made.

Moreover, previous studies have shown a significantly higher percentage of illiteracy and poverty in black and Hispanic populations. While the gap in the rates at which African-Americans and whites complete high school is narrowing, in the year 2000 there was still a 7 percent gap between them.

As far as the Scholastic Assessment Test (SAT) score averages, there are also significant differences between whites and other minorities. In the year 1999-2000, Asian Americans led the pack in mathematics with 565/verbal, 499; followed by whites, math, 530/verbal, 528; then Hispanics, math, 467/verbal, 461; and finally blacks, math, 426/verbal, 434. (5)

And since it has been said that a picture is worth a thousand words, let us look now at Table 2 (below) and compare such important factors as median household income and the level of poverty in the two sets of states.

Suffice to say, when it comes to median household income, the “high level of juvenile violence” states ($26,234) average only two-thirds that of the “low levels of juvenile violence” states ($39,127).

As seen on the same table, the lower median household income corresponds with the lower socioeconomic status of the “high levels of juvenile violence” states (percent of persons below the poverty level,18.9 percent vs. 10 percent). (6)

Discussion

And yes, the death of any child by any cause is a tragedy. I, too, decry the high levels of violence in our society. As a neurosurgeon who has spent incalculable hours in the middle of the night treating victims of gunshot wounds, I also deplore the level of violence and crime in America ­ but we must have the moral courage to find the truth and recognize the fact that there is another side to the story that is seldom promulgated by public health researchers and disseminated in medical literature.

We now know, for example, that the defensive uses of firearms by lawful citizens, at up to 2.5 million per year, dwarf the offensive gun uses by criminals. Between 25 and 75 lives are saved by a gun for every one life lost to a gun, whether by an unintentional shooting, homicide or suicide.

Medical costs saved by guns in the hands of law-abiding citizens are 15 times greater than costs incurred by criminal uses of firearms. Gun safety programs for children, particularly the NRA’s Eddie Eagle program, have resulted in a steady decline in unintentional injuries over the last several decades.

The number of guns in the civilian U.S. population has been increasing steadily for decades, and yet the number of fatal gun accidents has been falling for as long as statistics have been compiled (since 1903). In 1945, there were 350,000 firearms and 18 fatal gun accidents per million Americans. By 1995, although the number of guns in the U.S. had more than doubled, incredibly, there were 850,000 firearms with only 6 fatalities per million Americans.

Presently, according to government figures, at least 45 percent to 50 percent of U.S. households have firearms in the home. And yet the latest U.S. statistics show that the rates of serious crimes, including homicides with firearms and aggravated assaults, have fallen to record 25-year lows ­ despite the fact that the number of guns in the hands of Americans have exceeded the 200 million mark, nearly 1 million firearms per million Americans.

Furthermore, we should ask ourselves why, at the international level, Switzerland, New Zealand and Israel with modest gun control laws (as opposed to the draconian gun control laws in most of the rest of the world) and relatively easy availability of firearms, have low rates of homicide and violence, with or without firearms.

Because those countries ­ like Japan, which, on the other hand, has stringent gun laws ­ have an intact family structure and cultural cohesion, and civility and discipline are traditionally preserved in the population.

Yes, we must lay the blame where it belongs for the persistence of violence in America, and it’s not in “easy gun availability” but in many factors, as we have seen, pertaining to the cycle of government dependency, broken families, the failure of public education and the cultural disintegration that has been taking place for decades, particularly in the poorer Southern states, and for which the federal government has been largely responsible.

And, incidentally, if I have not been more pointed in addressing specific errors in my criticisms of this study by Miller et al., it’s because the raw, primary data that the authors used in formulating their conclusions have not yet become available for public review. (7)

One critic who has already raised pointed objections to specific errors in methodology leading to faulty conclusions in the study is Roger Schlafly, Ph.D., a concerned citizen and mathematician. He wrote me about this study as well as an article published in The Economist, “Bang, Bang, You’re Dead” (March 2, 2002), vaunting the study of Miller et al.

Schlafly states that in reality the highest childhood firearm death rates are in three rural states ­ Alaska, Montana and Idaho ­ with much hunting but less accessible emergency care. He holds that the study actually used “adult fatality rates to estimate firearm ownership levels so that the only insight gained by the study is that child firearm fatalities are correlated with adult gun fatalities.”

Schlafly also points out that although the authors mention gun survey data, Miller et al. instead used a modified “Cook’s index” ­ an index that they assert excludes suicides and homicides among children 0-19 years of age from the calculations in the study and that they affirm is highly correlated with gun availability measures.

The study thus ranked the states not only by “Cook’s Index,” the supposed proxy for firearm availability, but also to select the five “high-gun” and five “low-gun” states.

Roger Schlafly remarks, “I am a little baffled as to why this story would be news [in The Economist]. Wouldn’t everyone expect that high-gun states would have more gun deaths than low-gun states? Isn’t that obvious whether you are pro-gun or anti-gun? States with more cars probably have more car accidents also. Also, if you are involved in an accident that wasn’t your fault, perhaps you need to try contacting a personal injury lawyer for information on your claim.

“The only thing I found surprising was that ’16 times.’ If one state had twice as many guns, then I’d expect about twice as many gun accidents. But why 16 times? I doubt that any state has 16 times as many guns (per capita) as another state. The study neglects to mention that the child firearm accidental death rate is on a long-term decline.

“During the 10-year study period, the annual rate declined from 0.69 to 0.31 per 100,000. (Based on the same CDC figures used by the study, available online.)”

Obviously, my own objections, with which Schlafly also agrees and which were described for the reader in this two-part critique, are more broad and encompassing and extend not only to flaws in specific assumptions and methodologies, but also to the overall design of the study and the misuse of epidemiological research.

Miller et al.‘s complex study using an even more complex methodology ­ i.e., “pooled, cross-sectional, time-series data from the fifty states over a one year period” and the so-called “Cook’s Index” as proxy for firearm availability ­ are elegant examples of what the eminent social scientist and physician Dr. Bruce G. Charlton, of the University of Newcastle upon Tyne, calls “statistical malpractice.” (8)

The peer review process at the Journal of Trauma has failed ­ its reviewers lost with the authors in their own numbers and over fecund statistics. Miller et al. have used statistical malpractice in their failed effort to blame firearms per se on the tragic loss of children’s lives without seriously considering the role of more important variables.

Despite the authors’ claim that “state level analyses were adjusted for state urbanization, poverty, and education levels,” they actually failed to do so because of the lack of control over the overwhelmingly complex, confounding factors. What about other variables, such as reading and television-watching habits in children, culture and ethnicity, that were not even considered by the investigators?

Errors of this type are referred to as “ecologic fallacy” in epidemiological studies, whereby complex analytical techniques are combined with large data sets involving general populations extending over long periods of time.

Statistics are not science. They may show correlations, but they cannot prove cause-and-effect relationships because of the large amounts of variance and myriad numbers of confounding factors. (9)

This type of statistical legerdemain to prove cause-and-effect relationships in science was indeed exposed years ago by Dr. Charlton, who warned other scientists:

“Minimization of bias is a matter of controlling and excluding extraneous causes from observation and experiment so that the cause of interest may be isolated and characterized. … Maximization of precision, on the other hand, is the attempt to reveal the true magnitude of a variable which is being obscured by the “noise” [from the] limitations of measurement.” (8)

Conclusion

The authors of this study have failed not only to exclude extraneous causes from their “pooled, cross-sectional, time-series data,” but also they failed to incorporate into their analysis and conclusions the “true magnitude of the variables” ­ poverty, level of education, ethnicity, urbanization, race, broken families, etc. ­ among the children in the several states.

In fact, in observing cultural differences affecting most markedly the Southern states and analyzing the findings of the Journal of Trauma study, the thoughtful critic and honest scholar could very well ask himself if indeed the South has fully recovered from the devastation of the Civil War and the Reconstruction that followed more than a century ago. (10)

Try as the authors did, statistical adjustments failed to remove the multiple confounding variables. Dr. Charlton’s admonition was forsaken in the conception and design of this study.

“As a result, medicine has been deluged with more or less uninterpretable ‘answers’ generated by heavyweight statistics operating on big databases of dubious validity. Such numerical manipulations cannot, in principle, do the work of hypothesis testing. Statistical analysis has expanded beyond its legitimate realm of activity. The seductive offer of precision without the need for understanding is a snare to the incautious because exactitude is so often mistaken for explanation.” (8)

In other words, this published gun (control) study is an example of politically driven propaganda, result-oriented research with preordained, biased conclusions, which can only be characterized as junk science. (11-13)

Still, if the editors of the Journal of Trauma had insisted in publishing it, they should have printed it with an editorial caveat that such “pooled, cross-sectional, timed-series data” studies, involving such large populations over a long period of time, are utterly unreliable, even more so than Cohort and Case Studies analyses, subject to ecologic fallacy limiting their value in science.

The editors of the Journal of Trauma have done a disservice to their readers, obfuscating rather than contributing to solving the problem of violence in our society.

faria-13wmaz-sml

— Miguel A. Faria, Jr., M.D. is a retired Clinical Professor of Neurosurgery and Adjunct Professor of Medical History at Mercer University School of Medicine. He is Associate Editor in Chief and World Affairs Editor of Surgical Neurology International. He served on the CDC’s Injury Research Grant Review Committee.

All DRGO articles by Miguel A. Faria, Jr., M.D.

References/Notes

1. Miller M, Azrael D, Hemenway D. Firearm availability and unintentional firearm deaths, suicide, and homicide among 5-14 year olds. J Trauma 2002;52(2):267-275.

2. New Harvard University study shows direct link between gun availability and gun death among children. Press release. Violence Policy Center, February 21, 2002, http://www.vpc.org/press/0202study.htm.

3. For example, Professor Walter Williams of George Mason University writing in a syndicated column, “An ugly conspiracy of silence,” in the summer of 1999 analyzed the U.S. Department of Justice’s National Crime Victimization Survey (NCVS) and found that in the category of interracial crimes (1997) there were “1,700,000 interracial crimes, of which 1,276,030 involved whites and blacks. In 90 percent of the cases, a white was the victim and a black was the perpetrator, while in 10 percent of the cases it was the reverse.” Williams added, “Regardless of race, criminal violence is despicable and deserving of condemnation. But far more destructive are the official and unofficial attempts to mislead and conceal.”

Books that discuss this theme from different perspectives are David Horowitz’s “Hating Whitey and Other Progressive Causes,” Dallas, Texas, Spence Publishing Co., 1999; and Richard J. Herrnstein and Charles Murray’s, “The Bell Curve: Intelligence and Class Structure in American Life, New York, N.Y., Free Press, 1996.

4. National Center for Education Statistics. NAEP ­ State Profiles. http://nced.ed.gov/naep3/states/profile.asp?state=.

5. National Center for Education Statistics. NCES Fast Facts. Scholastic Assessment Test (SAT) scores. http://nced.ed.gov/fastfacts/display.asp?id=53.

6. National Center for Education Statistics. NCES Fast Facts. Table 2: Household income and poverty rates, by state: 1990, 1994, and 1995. http://nces.ed.gov/pubs98/98018/Tab2.prn.

7. Open-data, public review policy of the Medical Sentinel of the Association of American Physicians and Surgeons (AAPS). Medical Sentinel 1999;4(6):193-195.

8. Charlton BC. Statistical malpractice. Journal of the Royal College of Physicians in London, March-April 1996, pp. 112-114.

9. Milloy, SJ. Junk Science Judo. Washington, DC, Cato Institute, 2001, pp. 54-97. See also Bennett JT., DiLorenzo TJ. From Pathology and Politics: Public Health in America. New Brunswick, NJ, Transaction Publishers, 2000, pp. 80-83, 135-141.

10. Clark TD, Kirwan AD. The South Since Appomattox. A Century of Regional Change. New York, NY, Oxford University Press, 1967.

11. Faria, Miguel A Jr. The perversion of science and medicine (Part I): On the Nature of Science; (Part II): Soviet science and gun control; (Part III): Public Health and Gun Control Research; (Part IV): The Battle Continues. Medical Sentinel 1997;2(2):46-53 and Medical Sentinel 1997;2(3):81-86.

12. Faria, Miguel A Jr. Public Health and Gun Control ­ A Review (Part I): The Benefits of Firearms and (Part II): Gun Violence and Constitutional Issues. Medical Sentinel 2001;6(1):11-18.

13. Kates, Don B., Schaffer, Henry E., Lattimer, John K., Murray, George B., Cassem, Edwin H. Guns and public health: epidemic of violence or pandemic of propaganda? Tennessee Law Review 1995;62:513-596.

All Rights Reserved © 2016 Miguel A. Faria, Jr., M.D.