JAMA Braga et al; Firearm Caliber Correlation with Fatality

[berettaprofessor]

I haven't seen this recent (July 2018) article discussed, but it is pertinent to a discussion often held here: How much gun do I need?

https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2688536

To boil it down to the essence, the authors, Anthony Braga and Philip Cook (whom most of you are familiar with from their previous "studies"), looked at 367 cases of fatal or nonfatal shootings in Boston between 2010 and 2014. Divided into small caliber (22, 25, 32) medium caliber (38, 380, 9mm) or large (.357, 40, 44, 45, 10mm, and 7.62X39), they found that there was no difference on caliber category with number of wounds, wound location, or circumstances, but that medium caliber were 2.25X more likely to be fatal than small caliber, and large caliber 4.5X more fatal than small caliber. Buried in the text is also the point that indoor shootings were 2.6X more likely to result in fatality compared to outdoor shootings.

Their conclusion: "Firearms caliber was associated with the likelihood of death from
gunshot wounds in criminal assault. Shootings with larger-caliber handguns were more deadly but
no more sustained or accurate than shootings with smaller-caliber handguns. This conclusion is of
direct relevance to the design of gun policy."

The worrisome part is the last sentence of that paragraph; suggesting in no uncertain terms that gun policies leading to banning larger caliber weapons will result in less fatalities. They state clearly in their results section that if the public were limited to 22, 25, or 32 caliber weapons, gun homicides would decrease by 39.5%. They also make the statement in their concluding paragraphs that "It is plausible that larger reductions (in fatality rate) would be associated with replacing all types of guns with knives or clubs." Gee, ya think?

So, yes, at least one study suggests that bigger is better for projectile diameter; with the caveat that your goal is to kill, not just wound and deter.

 

[boom boom]

Knowing those two, it is a garbage study. One, their method is inappropriate for the question that they are asking.

Read it, decent dataset but crappy results and methods. No controls for bullet type, medical condition of the victim (old, young, female, male) by caliber type, shooter skill, location of wound, number of wounds, CNS hit versus vascular, time to treatment, etc. Energy levels between the rounds also plays a part--from examining a bullet--one cannot always discern whether a .357 or a .38 fired the bullet. Similarly on a .40 S&W versus a 10mm--it is the same caliber bullet. The grouping then of high powered .357 and 10mm with the .38 special or even the old .38 S&W into a "medium caliber". Distance from the shooter is also not reported.

The most ridiculous part was including a rifle round, 7.62x39 with medium caliber handgun rounds. Even weak rifle (and some handgun rounds shot through rifles such as .44 magnum) rounds have substantially more energy via velocity than most handgun rounds and will demonstrate increased mortality. If they had included shotguns, I suspect that the "large caliber" would have had even more lethality.

Omitted variables in a research design skew the coefficients of other predictive variable in an unpredictable fashion. That is made even worse by the failure of the model to predict most of the variance. Most of their table 3 model is garbage with very few variables demonstrating statistical significance and pseudo r's of about 3 percent and 6 percent explanatory value. If you look at the pseudo R2 in table 4, it is predicting about 7.5 percent in the variance regarding fatalities. But, with omitted variables which are admitted to in the study if you read carefully enough--e.g. skill, etc., the coefficient values of their explanatory variables are dubious.

There is no mention of collinearity checks on the variables and using "robust standard errors" is not a substitute for poor methodology.

So, someone shot in the head or central chest has a greater likelihood of dying, someone shot with a rifle has a greater chance of dying, and larger and faster cartridges are slightly more likely to cause death. Ho hum.

 

[JONWILL]

I bet this was done with money from a government grant. Interesting to find out who approved the funding.

 

[AlexanderA]

The demonstrators in front of the NRA Headquarters in Fairfax, Virginia, on Saturday, were demanding (in addition to their usual demands) a ban on all guns greater than .30 caliber. (This according to the Washington Post.) Caliber restrictions appear to be a new direction for the antigunners. The practical effect of such a thing would be to ban most handguns and shotguns. Of course they don't say that.

 

[JohnKSa]

Any differentiation of the data due to bullet types (i.e. expanding/non-expanding)? Not that I could see. Either they neglected it or they just didn't have the information.

No rationale on why they broke the calibers groupings where they did or why they included one rifle shooting.

They included calibers for which they had very sparse data--e.g. 3 of the calibers in the "large" group accounted for a total of only 6 shootings.

The caliber groupings and the inclusion of the calibers for which they had relatively sparse data definitely affected the results. For example, comparing the 9mm which was, by far, the biggest contributor in their "medium" group to the combination of the .40 and the .45 which were the main contributors in their "large" group resulted in a difference of only about 4% in lethality. Grouping they way they did and including the calibers with much more sparse data resulted in a lethality difference between their "medium" and "large" groups of about 18%.

I'm not saying that their groupings are wrong, just pointing that by regrouping things, or only focusing on calibers with significant numbers of shootings can give very different results and that means those two aspects of the study need to be examined very closely to insure they're not skewing the results.

Another thing I didn't like was that they ignored the fact that, on average, the fatal shootings involved 1.7 more shots fired than the non-fatal shootings. At least it was possible to note this in their data even though they ignored the effect.

Along the same lines, they neglected the fact that the guns in their "small" group have, on average, much smaller capacities than the guns in their "medium" group. In this case they did not provide the average number of shots fired, broken out by fatal/non-fatal shootings for each caliber group so there was no way to even see if there was a difference in the shots fired although common sense and basic firearms knowledge strongly suggests that there would be.

It's pretty obvious that higher capacity/more shots fired is going to have an impact on lethality but although their data shows about 2 more shots, on average, fired for fatal shootings than non-fatal and it's clear that the guns in the "medium" and "large" group will have, on average, significantly higher capacities than the guns in the "small" group, they still conclude that the lethality difference is primarily due to caliber.

It's also probably true that the guns in the "medium" and "large" groups are, on average, considerably more shootable than the guns in the "small" group but this is also not taken into account, or even mentioned.

I think that the data (and common sense) does support the idea that people shot with "more powerful" calibers are more likely to die as a result, but I think a more carefully done study wouldn't show anywhere near the differences that they "observed" in their conclusions.

Ok, all of that is playing their game; just pointing out that they should follow the rules of the game more closely.

It's critical to keep in mind that when looking at a study, or answering a question, it's important not to get so bogged down in the details that you lose track of what's really important.

So let's take a step back.

Should we be appalled/concerned/worried that more powerful calibers are more likely to be lethal? No, of course not.

We choose more powerful handguns (e.g. 9mm instead of 22LR) because they are more likely to be effective in self-defense situations. In other words, they are more useful/effective for saving innocent lives of people attacked by violent criminals. In that context, lethality isn't really a major factor, but to the extent that it figures into the equation at all, it's more likely to be desirable than undesirable since there is almost certainly strong correlation between the likelihood of a handgun wound to be lethal and its ability to stop an attacker.

 

[ATLDave]

On the other hand, this does lend a bit of credence to a conjecture of mine: magazine capacity limits may lead to an INCREASE in fatality rates if people (rationally) shift to larger calibers to make up for a loss of capacity.

I also have to laugh at the ban-big-bore stuff. Many anti-gunners have been decrying the “evil” .223 round for its lethality combined with low recoil - allegedly making it especially dangerous in mass shootings. They’re just all over the place.

 

[Old Guy]

Stabbings, as in Jailhouse assaults, tend to be many puncture wounds, ending in lots of fatalities?
If the criminal attack is performed with the carotid artery cut? Fatality more or less guaranteed!
Not sure were the knifing figures come from? Having been stabbed twice in my days on the door of Clubs in Liverpool UK, and am still here? Possibly being able to move freely, in the doorway, and in the street contributed to my survival. Plus I could fight! As an 82-year-old Grandad, now carrying a Glock 19 every day (plus never being out after 10 pm!) being in any kind of fight? Is somewhat a non-starter.

 

[somethingbenign]

Didn't even think of this study being used for control arguments. I also don't know who these authors are so I'm sure I'm missing some background information. What handgun rounds would even be left if limited to 30 cal? All I can think of is 22lr and 25acp, both of which would be amusing to see how many you could fit in a full frame gun.

 

[ATLDave]

5.7 would be left. Of course, that’s a round that may have more ability to penetrate typical LE vests... not what the controllers have in mind. Because they don’t actually know about guns, they constantly propose things that would have the opposite effect of what they envision.

 

[lemaymiami]

Any study that doesn't include the actual distance between the weapon and the victim of any shooting is pretty much useless in my experience... and the actual results on the street of a shooting incident where someone was hit by round(s) is so random as to be difficult to quantify. I was personally on the scene of injuries that should have been fatal - but the victim survived in good shape, and incidents where what looked like minor injuries killed the victim (or offender)... Very glad those days are long behind me.

Of course those that oppose weapons or self-defense in any form will use whatever studies support their assertions... surprise, surprise....

 

[boom boom]

Cook is a notorious "public health" researcher who makes his career on anti-gun studies. John Lott has debunked several of Cook's studies.

My guess is that this was financed via Bloomberg, the Joyce Foundation, or some other sort of group. Public grants are normally acknowledged in the paper and data would have to be made accessible to other researchers in time.

 

[Double Naught Spy]

Cook is a notorious "public health" researcher who makes his career on anti-gun studies. John Lott has debunked several of Cook's studies.

Cooke may be, but saying that John Lott has debunked several of Cook's studies is about like to pot calling the kettle black. John Lott, himself, has made some dubious claims such as "more guns less crime."

 

[denton]

Just a quick scan through the stats.... The only place they give P values is in the Chi Squared test. Those values are totally unimpressive. The conclusion a fair researcher would reach is, almost none of these variables have an impact that can be distinguished from random noise.

 

[edwardware]

Buried in the text is also the point that indoor shootings were 2.6X more likely to result in fatality compared to outdoor shootings.

Oh, Oh, quick! Ban indoor guns!

Next up, a study correlating fork tine length with childhood obesity.

. . . a fair researcher. . .

Not a one of those in sight.

 

[boom boom]

Cooke may be, but saying that John Lott has debunked several of Cook's studies is about like to pot calling the kettle black. John Lott, himself, has made some dubious claims such as "more guns less crime."

Actually, to the best of the limits of the data, Lott has a pretty good track record on research. He had to as he is definitely cutting against the grain in academia. I have actually read Lott's studies, including his most recent book, and understand the methodology and data limitations of his studies. I also am pretty acquainted with one of his co-authors who is a straight up excellent econometrician outside of his work with Lott.

It is always difficult because of the ecological fallacy to tie aggregate data to individual level actions such as crimes. However, Lott's aggregate analyses tracks with Gary Kleck's work who does use individual level data to determine instances of Defensive Gun Use. Kleck revisited his original works recently with new data and indicates if anything, that he undershot his original estimates. Even Kleck's enemies admitted that roughly 800,000 DGU's occur annually while Kleck's original and revised estimates put it well over 2,000,000. As to qualify that you must have a gun for a defensive gun use, those states with more permissible gun laws have more defensive gun uses that actually prevent crimes.

Lott buttresses that claim and if you like you can download his data and produce your own obviously superior statistical analysis.

BTW, you should know that there is a serious crisis in medical research, psychology, sociology, etc. in data, methodology, replicability, and ethics. Famous studies such as the prison guard study, the wild boy's study, Margaret Meade's study of Polynesian culture, repeated failures in studies on nutrition, drug studies, Freud, even Milgram's experiment have been shown in hindsight to be littered with dubious ethics, poor controls on data, coaching respondents, and so forth. The only way to advance in science is to use openly available data and methods and not allow "secret sauce" to be added by authors. Lott does this and while you apparently dislike his implications and research, you are free to apply his data and methods to analyze it to your heart's content.

Philip Cook's analysis in this instance and in other instances where I have read his papers fall far short of adequate persuasion to someone who knows how to conduct research. Mis-specifying his statistical model by failure to add known control models in the research piece linked above is a grave mistake that makes any attribution of causal factors via coefficients suspect. Given the weakness in explaining variance overall coupled with the inability to attribute factors actually explain the variance means it is garbage fit only for media and anti-gun group consumption. Marshall and Sanow, by way of contrast, while having problems with the data, did straight forwardly acknowledge problems in their research.

 

[CDW4ME]

They might have an aneurysm if they saw expanded 45 HST- about the size of a quarter.
9mm HST on dimes, 45 HST on quarters, for size perspective.

 

[boom boom]

They might have an aneurysm if they saw expanded 45 HST- about the size of a quarter.
9mm HST on dimes, 45 HST on quarters, for size perspective.
View attachment 799329

Note that these expanded bullets would be treated the same as ball 9mm or 45 ACP ammo in lethality in the research above.

 

[hso]

if the public were limited to 22, 25, or 32 caliber weapons, gun homicides would decrease by 39.5%

Except for the ignored fact that criminals are already breaking the most sacrosanct law in attempting to murder and that they'd be undeterred by caliber restrictions.

 

[rskent]

So, my understanding of the study results are: That bigger bullets make bigger holes in people.

And that blood runs out faster from bigger holes. Duh.

 

[Speedo66]

Do they say how long it took a person to die? Lethality means nothing if it takes hours or days to happen.

The most important question is not whether the victim died, but whether they were stopped from continuing what they were doing, i.e., did the bullet instantly stop the threat.

 

[jmorris]

They state clearly in their results section that if the public were limited to 22, 25, or 32 caliber weapons, gun homicides would decrease by 39.5%.

But that’s moronic....

 

[boom boom]

Look, what their tables show is the mean estimate of the coefficient's effect. When actually setting up reputable research, the variation would demonstrate a confidence interval for the potential effect (similar to polling where the mean is cited followed by the +- error and confidence interval).

However, the overall explanatory model in Table 4 explains about 7.5 percent of the variations in mortality. Most of the control variables do not make much sense--e.g. race differential on mortality would indicate that people differed on mortality from handgun wounds which does not make sense logically. Instead, the proper variable would be weight and height. Same thing on indoor variable which is probably collinear (correlated with) with the unspecified (not included) distance variable of shots fired. Closer, it is easier to obtain good hits than farther, etc. Time of day or night also makes a difference on shooter accuracy which is not included in the model. Now, calculating t values for the coefficients of small and medium calibers indicate 3.20 and 4.57 which is decent for a one-tailed test and the dataset size, however, these are the only variables included in this table.

Absent are the variables of distance, time to medical treatment, access to trauma centers, shot placement, type of bullet, etc. mentioned above which are present in the Table 3 model. The statistical models in table 3 fail to explain much of anything and demonstrate a very poor fit to boot. Table 2 is simply descriptive statistics.

One of the test of good research is that any result that does not employ past research on lethality into its model is junk. We know from both animal testing, gelatin, and morgue examinations certain things about lethality of projectiles. None of this research was presented in this paper. The arbitrary .32 caliber and .22 caliber distinction allows a 5.56 round to be included under small caliber as well as the .32 H&R and .32 Magnum. I guess if you shot with a .30 caliber rifle round or worse yet a .270, that these would count as small caliber. Bullet placement which relates to distance, training, and other circumstances is also of vital importance. For a medical journal, the absence of any wound analysis strikes me as remarkably bad.

There is an old maxim in mathematics that your findings are governed by the least precise variable in your analysis. That means that the fake certainty of their odds ratio analysis conceals that other factors not specified (mentioned) in the model control over 90 percent of the variance from whether someone lives or dies from a gunshot wound.

 

[berettaprofessor]

Philip Cook's analysis in this instance and in other instances where I have read his papers fall far short of adequate persuasion to someone who knows how to conduct research.

Yes, but the problem is that the loudest mouthpieces for public policy change (i.e. politicians during election cycles and/or anti-gun activists working from traumatic events or emotional loss) don't either understand how to conduct research or they don't care how it was conducted if the published conclusions supports their desired change.

 

[BPchic]

A long ago study showed that knife and handgun wounds have similar fatality rates. It's the size of the hole and damage caused. Obviously rifles and shotguns have higher rates (if you leave out small rifles). It all relates to the size of the hole. A large knife (or a sweeping cut or many cuts from a small knife) should be more lethal than a simple stab wound from a small knife. I have seen no study even defining large vs. small knives, but I'd be glad to do one if someone will sufficiently fund it.