Why gun science is such a murky topic

[Heffay]

http://boingboing.net/2013/02/26/firear ... e-mis.html

I know a lot of people are beholden to their views, and discount studies that oppose their positions while supporting studies that do. Which is human nature (confirmation bias). However, with gun research there seems to be a LOT of conflicting data. This article discusses why this happens, and it appears to be mostly related to the data being crap. Different standards, unclear standards, etc, etc.

Anyway, I found it interesting reading.

[Quasar]

Objective read, thanks for widening the debate and keeping things interesting.

[antimatter]

These sorts of problems aren't restricted to firearms. One of my other hobbies (passion, really) is riding motorcycles. When I started riding in the late '70's, the only definitive research done on rider safety had been done by a guy named Harry Hurt (funny, I know). Basically, Dr. Hurt complied data on 900 motorcycle crashes, and used that information to extrapolate a lot of the data that was used to form the training given to new riders by the Motorcycle Safety Foundation (MSF). Blah, blah, I could go on, but the point is that even to this day there isn't really a good study on why people on motorcycles crash, and ways to prevent it. Heck, the same issue is pretty murky for automobiles, so much so that organizations like the Insurance Institute for Highway Safety (IIHS) prefers to focus on stuff like ABS, stability control, and leave the human variable out of the equation. If you're wondering why there are so many bad drivers out there, it's because the government's focus on safety has relied on studying non-human factors. So driver's ed is essentially a joke, and most drivers never revisit it or get any up to date training. And a great deal of this is because no-one wants to pay for an in-depth study that would used standardized data to figure out the best ways to train driver how not to get into accidents.

Usually, when money and politics gets involved in science you have a bad result.

[jdege]

The primary reason the science is so weak is that it's being done by Public Health professionals, rather than actual scientists.

We did have some actual science done in the area, 20+ years ago, by criminologists and sociologists. And they reached a consensus:

http://www.guncite.com/journals/tennmed.html
Scholars engaged in serious criminological research into "gun control" have found themselves forced, often very reluctantly, into four largely negative propositions. First, there is no persuasive evidence that gun ownership causes ordinary, responsible, law abiding adults to murder or engage in any other criminal behavior--though guns can facilitate crime by those who were independently inclined toward it. Second, the value of firearms in defending victims has been greatly underestimated. Third, gun controls are innately very difficult to enforce.

The difficulty of enforcement crucially undercuts the violence-reductive potential of gun laws. Unfortunately, an almost perfect inverse correlation exists between those who are affected by gun laws, particularly bans, and those whom enforcement should affect. Those easiest to disarm are the responsible and law abiding citizens whose guns represent no meaningful social problem. Irresponsible and criminal owners, whose gun possession creates or exacerbates so many social ills, are the ones most difficult to disarm. A leading English analyst's pessimistic view has been summarized as follows: "[I]n any society the number of guns always suffices to arm the few who want to obtain and use them illegally ...."

Therefore, the fourth conclusion criminological research and analysis forces on scholars is that while controls carefully targeted only at the criminal and irresponsible have a place in crime-reduction strategy, the capacity of any type of gun law to reduce dangerous behavior can never be more than marginal.

Now if the gun control proponents had been honestly interested in the truth, they would have given up their cause as counterproductive, when these findings became clear. But they weren't, and proceeded to ignored the findings of the experts and the research that they had funded, and moved over to the public health sphere.

Why? Because the standards for peer review in the public health journals is truly abysmal. And because most practitioners in public health think that epidemiological studies can provide proof - which they simply cannot. Hence Kellerman and the rest - bogus junk, masked in the trappings of science, published by journals that should have known better.

[Heffay]

The primary reason the science is so weak is that it's being done by Public Health professionals, rather than actual scientists.

We did have some actual science done in the area, 20+ years ago, by criminologists and sociologists. And they reached a consensus:

This, ladies and gentlemen, is what we call "confirmation bias".

[Snowgun]

The data is really crappy, there are so many confounding factors and covariates that it's hard to have a definite conclusion.

However, One thing i've seen is that John Lott's findings have in general been checked quite heavily, especially since he makes the data set available, by many famous statisticians, including Steven Levitt. While there is NOT consensus on whether the data show that guns decrease crime or have no effect, there IS a overwhelming majority of statisticians that conclude that it doesn't increase crime (with of course a few that go great lengths to twist the data to show that it does).

In other findings, it has been shown beyond a doubt that Glocks suck in comparison to 1911's.

[Holland&Holland]

I was very disappointed with this link. The thread title promised gun science, but not one mention of velocity or ballistic coefficients.

[Heffay]

In other findings, it has been shown beyond a doubt that Glocks suck in comparison to 1911's.

YOU TAKE THAT BACK!

[jdege]

The primary reason the science is so weak is that it's being done by Public Health professionals, rather than actual scientists.

We did have some actual science done in the area, 20+ years ago, by criminologists and sociologists. And they reached a consensus:

This, ladies and gentlemen, is what we call "confirmation bias".

When the NIJ hired two of the most highly-respected sociologists in the country - James Wright and Peter Rossi - to do the definitive study on the effectiveness of gun control, they expected to them to produce results that would confirm the need for increased gun control. After all, both had written approvingly of gun control in non-academic publications. Unfortunately, they actually did the research and were guided by the results, rather than politics.

They found that there was no evidence that any gun control measure had had any positive effect, anywhere, ever.

Which should have come as no surprise. As it's the result every other honest researcher has come to.

Up until about 1976 or so, there was little reliable scholarly information on the link between violence and weaponry. Consequently, everyone, scholars included, was free to believe whatever they liked about guns and gun control. There was no scientific evidence to interfere with the free play of personal bias. It was easy to be a "true believer" in the advisability of gun control and the uniformly detrimental effects of gun availability (or the opposite positions) because there was so little relevant information to shake one's faith. When I began my research on guns in 1976, like most academics, I was a believer in the "anti-gun" thesis, i.e. the idea that gun availability has a net positive effect on the frequency and/or seriousness of violent acts. It seemed then like self-evident common sense which hardly needed to be empirically tested. However, as a modest body of reliable evidence (and an enormous body of not-so-reliable evidence) accumulated, many of the most able specialists in this area shifted from the "anti-gun" position to a more skeptical stance, in which it was negatively argued that the best available evidence does not convincingly or consistently support the anti-gun position. This is not the same as saying we know the anti-gun position to be wrong, but rather that there is no strong case for it being correct. The most prominent representatives of the skeptic position would be James Wright and Peter Rossi, authors of the best scholarly review of the literature.

[Subsequent research] has caused me to move beyond even the skeptic position. I now believe that the best currently available evidence, imperfect though it is (and must always be), indicates that general gun availability has no measurable net positive effect on rates of homicide, suicide, robbery, assault, rape, or burglary in the U[nited] S[tates]. This is not the same as saying gun availability has no effects on violence -- it has many effects on the likelihood of attack, injury, death, and crime completion, but these effects work in both violence-increasing and violence-decreasing directions, with the effects largely canceling out. For example, when aggressors have guns, they are (1) less likely to physically attack their victims, (2) less likely to injure the victim given an attack, but (3) more likely to kill the victim, given an injury. Further, when victims have guns, it is less likely aggressors will attack or injure them and less likely they will lose property in a robbery. At the aggregate level, in both the best available time series and cross-sectional studies, the overall net effect of gun availability on total rates of violence is not significantly different from zero. The positive associations often found between aggregate levels of violence and gun ownership appear to be primarily due to violence increasing gun ownership, rather than the reverse. Gun availability does affect the rates of gun violence (e.g. the gun homicide rate, gun suicide rate, gun robbery rate) and the fraction of violent acts which involve guns (e.g. the percent of homicides, suicides or robberies committed with guns); it just does not affect total rates of violence (total homicide rate, total suicide rate, total robbery rate, etc.).

[AFTERMATH]

Statistics show that 73% of statistics are made up.