[Editor: With this article, we introduce the work of our Publication Review team of medical scientists and statisticians who add their expertise to countering the ‘political’ science endemic to anti-gun ‘research’.]
We have blogged about Boston University anti-gun advocacy researcher Bindu Kalesan’s commitment to attacking the very culture of American gun owners. Now, with five coauthors and a little help from her friends at Michael Bloomberg’s billionaire-fueled gun control group Everytown for Gun Safety, Kalesan has crafted a new political message dressed up as legitimate science. And she has published it in what has become a mouthpiece for gun-hating American academics, the British Medical Journal publication Injury Prevention.
We’ll have more to say about Kalesan in Thursday’s blog entry, but today let’s do a surgical dissection of her article, “School Shootings during 2013-2015 in the USA” (full text is behind a paywall). Kalesan and her coauthors used a list of media reports of school shootings compared with a list prepared by “Everytown USA”, a term that yields Google results only about Everytown for Gun Safety.
Kalesan compared 154 of these reports, using cases in which “a firearm was discharged inside a school building or on school or campus grounds…” Schools included elementary through college levels, according to the list in Supplementary Appendix 2. In all, 57 deaths were reported (excluding the shooter) and 134 nonfatal injuries.
Stay with me, because here’s where she goes off the rails, reality-wise. Using a variety of public data sources Kalesan gathered information about state laws requiring background checks for handgun purchases, mental health expenditures by the various states, state money spent on K-12 education, the percentage of population living in cities in each state, and rates of gun ownership. Still with me?
Then using an extremely complex and opaque statistical analysis, Kalesan and her coauthors came to this conclusion:
School shootings are less likely in states with background check laws, higher mental health and K-12 education expenditures, and with greater percent urban population.
But DRGO’s Publication Review Team reviewed Kalesan et al.’s own data and found far different results using much simpler data analysis. The following is what they found using the study’s list “Incidents of school shootings from 1 January 2013 to 31 December 2015 (Table 1).”
The major premise of Kalesan’s article is the third finding: “Third, we found that states with background check laws for firearms and ammunition purchase, higher capita MHE [mental health expenditure] and KEE [K-12 education expenditure], and higher per cent urban population had lower school shooting incidence rates.”
This conclusion is possible only through cherry picking the data. If we sort the raw data by number of deaths, the states with the leading body counts are Oregon (11 victims), California (7 victims), and Washington (5 victims). These are all liberal states which have the very background check laws that the authors claim reduce school shootings.
The raw data of this article indeed shows something completely opposite from what the authors claim—a strong statistical correlation between restrictions on gun rights and deaths in school shootings.
The constitutional carry states (Alaska, Arizona, Arkansas, Idaho, Kansas, Maine, Mississippi, Vermont, West Virginia, Wyoming) display as a group these averages, with confidence intervals specified after each average. These are strongly statistically below average:
- Incidents: 1 +/- 0.866
- Dead shooter: 0.375 +/- 0.696
- Dead victim: 0.375 +/- 0.696
- Injured victim: 0.625 +/- 1.111
Compare with the 9 states with the most restrictive gun laws (California, Connecticut, Illinois, Hawaii, Rhode Island, New York, Maryland, Massachusetts, New Jersey):
- Shootings: 1.33 +/- 2.05
- Dead shooter: 0.11 +/- 0.31
- Dead victims: 0.89 +/- 2.18
- Injured victims: 67 +/- 3.77
19 states require background checks on private sales of handguns (CA, CO, CT, DE, HI, IL, IA, MD, MA, MI, NE, NV, NJ, NY, NC, OR, PA, RI, WA). Averages are:
- Shootings: 2.63 +/- 2.87
- Dead shooter: 0.52 +/- 0.82
- Dead victims: 1.58 +/- 2.87
- Injured victims: 2.94 +/- 3.63
To summarize, even if we were to take the authors’ raw data at face value it does not support their conclusion. Constitutional carry states have significantly below average school shootings. In states with restrictive gun rights, shootings are slightly above constitutional carry states, but it is much more likely that an innocent will die or be injured than in a constitutional carry state. States that require background checks on private sales are close to national average on shootings and dead shooter, but show a significant increase in dead and injured victims.
Further, data is missing in at least two states the authors cite:
- Their data for Massachusetts shows one incident, with zero victims killed or injured. This is simply not true. On April 18, 2013 an MIT campus police officer was shot multiple times and killed by the Tsarnaev brothers. Tamerlan Tsarnaev died in a shootout with police soon thereafter.
- Their data for Connecticut shows zero incidents. But on January 13, 2014 in New Haven a 14 year old boy was shot outside a basketball game at the Hillhouse High School athletic facility, suffering wounds in his hand and leg.
For this blog entry DRGO’s Publication Review Team also looked at school shootings from another angle—whether school shootings were less frequent in states with restrictive college campus carry laws.
The Armed Campuses website shows 6 states with the least restrictive campus carry laws. This includes the red states on the map (TX, CO, UT, ID, and TN) as well as one of the grey states (MS). The other 44 states demonstrate a range of restrictions. Some allow concealed guns on college campuses, but schools have veto power and no schools have yet allowed carry (grey states, except MS); concealed guns are prohibited on campus by law (green states); concealed guns are only allowed in locked cars (orange states); or laws leave it up to the schools to decide, and none of these have yet allowed carry (yellow). So essentially, 6 states allow campus carry and 44 effectively ban it.
School shooting data, including deaths and injuries, from 2010-present are taken from Wikipedia. The 6 least restrictive states have had 0 shootings, and thus 0 shooting-related deaths or injuries, since they enacted their current carry laws. The time of enactment was different for each state: TX since Aug 1, 2016; CO and UT for the entire period (2010-2016); ID since 7/1/14; TN since 7/1/16; and MS since 2011. Looked at another way, expressing the denominator as state-years (e.g., for CO it is 1 state x 7 years, or 7 state-years) there were 0 shootings over 23.16 state-years.
On the other hand, looking at the remaining 44 states from 2010-2016 (as well as the above 6 states for the period before they liberalized campus carry), there were 41 campus shootings that produced 52 deaths and 66 injuries over a period of 326.84 state-years. The incidence rate then is 41 shootings/326.84 state years (0.12), 52 deaths (0.16), and 66 injuries (0.21).
So this straightforward analysis of school shootings shows a dramatically smaller number of school shooting deaths—zero, in fact—in states that allow campus carry compared to states that restrict it. Specifically, there were 0 shooting-related deaths per 23 state-years in the states that allowed campus carry vs. 0.16 in the states that did not. Applying Student’s t-test shows this comparison to be highly significant, with a p value of 0.0006.
Tellingly, the paper’s discussion section is chock full of caveats and hedges, e.g., “The negative association between per capita MHE [mental health expenditures] and incidence of school shooting [sic]…needs to be interpreted cautiously in light of our study’s limitations, noted below, and seen as preliminary.” But this warning by the authors didn’t stop Business Insider from headlining this dicey finding as fact.
The authors go on to note “five main limitations” in their study:
1) they “used exclusively media reports” as the source of school shooting data, which they admit are often incomplete. Indeed, the authors’ list does not line up with Wikipedia’s list of school shootings;
2) the authors say they were unable “to identify the mental health status of the perpetrators”, making their assertions about mental health funding influencing shootings irrelevant;
3) the relatively small number of shootings weakens their statistical analysis; and,
4) shootings resulting in no fatalities “may be more likely to go unreported by news agencies”, thus skewing the results.
The authors themselves describe a fifth limitation of their study, an astonishing confession that their claims about mental health and K-12 education spending are practically worthless: “Fifth, to our knowledge there are no studies that have validated the mental health and education data used in our study.”
This goes against all established scientific practice. Researchers have a professional obligation to validate the sources of their data before using it to prove their cases.
Kalesan and her coauthors only show, one more time, that if you torture the data long enough, it will tell you anything. The editors at Injury Prevention share the blame; they should never have allowed this paper to be published.
Few mainstream media reporters are capable of critically assessing the claims of anti-gun rights advocacy researchers like Bindu Kalesan. And even fewer are inclined to, since they enthusiastically endorse her message—guns are evil, gun owners are ignorant at best, and gun ownership should be banned. So CNN, AOL News, and a host of other old media outlets willingly and uncritically reported the authors’ claims. Mission accomplished.
If Kalesan and the editors of medical journals can’t be trusted to give us the truth about guns, what is to be done? DRGO will look at the big picture of gun prohibition advocacy research in the next blog entry. Stay tuned.
—Timothy Wheeler, MD is director of Doctors for Responsible Gun Ownership, a project of the Second Amendment Foundation.