Well you got something alight, but the last quoted author was from 2004, demanding that the authors of 2009 had look at him; perhaps, but that is not obligatory. The 2009 researchers also put a lot of caveats in their research implying, as good researchers should, that indeed more research was needed.
And of course this is ongoing. What it has to be mentioned is that if it was as flawed as the critics report then you know what should had taken place; and considering the resources the gun lobby if the research was as flawed: there is a precedent in cases of mistakes and flawed research papers published by climate change deniers. Papers that have been criticised so effectively that it leads to either the removal of the paper by the science journal or the editors quitting or being fired for incompetence.
What it should be noticed here is that I have to say that IMHO the lack of a concerted effort to do this (demand removal of flawed papers or retractions or have editors inconvenienced) shows to me that many of the people that do check on the research do not see the flaws as the gun proponents wants them.
And of course if the criticism had carried a lot of water it would had told other researchers to shy away from using the paper. But it what many of the critics ignore is that even if one can say that a paper has flaws, there is information to be used, as seen in a more recent paper:
In the paper the researchers approach this from a mathematical angle, they concluded that:
While the metrics they used make a better case for controlling guns (or even banning them, but I do not see that in my lifetime) more research is needed to see if it could go the other way, but the main reason why I looked at subsequent research was to see how other researchers see that 2009 paper that so ruffles the OP:
And I had to miss the link to the last quoted paper:
Dependence of the Firearm-Related Homicide Rate on Gun Availability: A Mathematical Analysis
Dominik Wodarz1,2,* and Natalia L. Komarova1,2
Derek Abbott, Editor
I dislike all studies that compare the increase, reduction, or possession of firearms with “firearm related” offenses. So what!? I like to see studies that compare those things with all violent crime or murders, etc. When a study shows that owning a gun in the home make people X% more likely to have a gun accident, I cringe. You needed a study for that? Or that an increase in guns results in more gun violence. But does it result in less overall violence?? I don’t know, but that’s what I’d like to see in a study. Gun bans result in a decrease of gun-related homicides? Really? That’s great, but what correlation does it have with all homicides? Maybe there is a decrease in homicides as a whole, due to the protections afforded by guns, but as a consequence, the gun-related murders are up. I think these kinds of studies are much more valuable than all the ones comparing gun bans to gun crimes. Compare them to the rates of all violent crimes.
It was about ‘how silly it was for scientists to investigate why and how many more Russians die younger for abusing vodka’.
Why would someone waste money in a study for what it is obvious or known? As I noted, and most also reported to that OP:
“Even with a little educated guessing I think that groups like insurance entities in Russia will be able to use that data to make better risk assessments and save money. The point is that regardless what one thinks that ‘some things are so obvious that they should not be investigated’, what happens is that studies are not done to benefit the preconceptions of armchair scientists; but are made also to help industry and government to quantify an issue so resources can be used better.”
So it is for something seemingly silly as looking at how often gun accidents happen at home. It is not only governments or people worrying about guns, but insurance companies and even gun owners that should not be in the dark about what to do to lower risks in those situations.
This page has a couple of illuminating charts. It shows that while the US is in the bottom half of overall crime rates, both violent and nonviolent, it has a hugely higher incidence of homicide.
I get it. I’m not saying that there isn’t valuable information to be obtained from learning about accidents rates, etc.
I’m referring to people with anti-gun agendas who quote results from studies such as, “People who own guns are X% more likely to commit suicide with a firearm.” “Countries with more guns have more gun crime.” Do they have more crime? Or just more gun crime. Would the gun owner have killed themselves by some other means if they didn’t own a firearm? Are the suicide rates higher among teens whose parents have guns in the home? Or is it just suicides-by-gun that are higher!?
And I’m not saying these types of studies don’t exist. I know there are studies that address these things. My comment was merely to suggest that these studies, and not the ones I described, are the ones that should be used when discussing gun control. It’s disingenuous and irrelevant to talk about reduction in gun crimes that will result in fewer guns. Yea, no shit.
Right. And those are the kinds of statistics that are relevant to gun-control debates. Please don’t get me wrong. I’m not trying to make a pro OR anti gun statement here. I’m just making a comment about particular type of study and/or statistic thrown around during gun-control debates. In fact, there is a decent example among your illuminating charts.
This chart for example, makes the illuminating claim that assaults in the US are more likely to involve guns than countries without so many guns. That’s exactly the type of pointless, obvious statistics that I’m talking about. I’m not saying that, taken as a whole, the paper doesn’t make a decent argument for greater control, etc. I’m just saying that it is grating to here people throw out, all by itself, that kind of statistic and act like it somehow shows that guns are a problem. Having more assaults involve firearms says nothing about the overall rates of assault in general. Maybe the rate of total assaults is drastically lower in the US.
Gun control discussions should, in my opinion, focus on overall crime not just gun-related crime. This is because one of the specific theories of pro-gun advocates is that having a gun makes someone safer and prevents crime. In their mind they will think that more guns on the streets and more people carrying guns will result in fewer people getting robbed or raped, etc. “Sure, more assaults involve guns in America,” but you haven’t disproven the idea that, “an armed population reduces the rates of assaults among that population.” In fact, your cite even mentions the lower rates of total assaults in the US, but explains it away as differences in reporting, etc.
So an armed population of 100,000 people, might have 3x as many gun-related assaults. But it might also have half as many assaults as other countries. Given armed country A and unarmed country B, both with 100,000 people. A has 100 assaults, and 30 gun-related assaults. Country B has 300 assaults, but only 10 gun-related assaults. With those numbers, Country A has 3x the number of gun-related assaults, but also 1/3 as many assaults overall. So which country has safer citizens?
Anyway, I think I’ve put way to much explanation into this. And people are going to think I’m trying to make some kind of pro-gun argument here. But I’m not. I’m just trying to point out that studies showing, “… that while the US is in the bottom half of overall crime rates, both violent and nonviolent, it has a hugely higher incidence of homicide” are so much more relevant and important in gun control debates than, “the US has X% more gun-related whatevers than these other countries”. If, instead, galen ubal’s quote from a study was that “…it has a hugely higher incidence of gun-related homicides”, it would have been the type of cringe worthy statistic I’m talking about. Instead, it actually references all homicides, and is therefore a useful, meaningful piece of information.
First - while interesting I’d like to leave more generalized information out of this thread, as well as other various studies. There are lots of different threads to discuss the general issues, but this one is about a specific study.
I’m not questioning that the authors performed mathematical calculations to come to their results. The ideal state would be to match the controls to the case examples exactly save for the variables trying to be studied. The more the control examples vary from the case, the more adjustments needed to control for those. Smallish differences are better than largish ones. But if the differences in population are very large, then larger adjustments are necessary to align them - to control for them. How large an adjustment is needed when the case examples varied by location (indoor/outdoor) by 900%? So yes, the authors probably made adjustments for this, but the larger the adjustment, the less accurate the predictive value the results would yield.
And this is just on the location criteria which I’ve focused on thus far. There’s plenty more shortcomings in this study.
Consistency in result is good. Of course, if a series of studies consistently used methodology that was poor, the result would be consistently poor - that doesn’t make the results accurate. So yes, this thread focuses on a single study because it focuses debate. General directionless discussion of firearms is available across the boards if that’s your fancy.
Yes - the case population had a much higher rate of criminal background. Another significant weakness. It’s as if being engaged in criminal enterprise wouldn’t be a characteristic to influence violent encounters.
I think it can be relevant. The authors starting point for case examples were those that were shot after being assaulted - victims of a crime. Guns aren’t exactly like a disease that can be treated the same way other diseases are. There may be overlap in analytical methods, but I wouldn’t go so far as to exclude an entire discipline.
Not everything is about climate science.
That’s all the time I have for now. I’ll get back with more on some of the other weaknesses in the study outside of the location aspect.
To my knowledge I haven’t seen this study before this thread. I viewed it noting some key data at the beginning, that 3,202 individuals were assaulted with guns over a period of three years, in Philadelphia, and their select group of 677 is overwhelmingly black, male, and outdoors at night in places with high drug trafficking.
The percentage given for gun possession is lower than I would have guessed, and the percentage of people shot reported as “professional” is higher than I would have guessed, and kids were excluded from the study, which suggested to me that the problem goes beyond gangs battling each other and into the general public. The general public still has to live in and traverse and do business in high-crime areas, and who there isn’t interested in self-protection?
The conclusions boil down to guns not being great defensive tools in places where guns and violence abound. By all means shed some light on this if you think it’s warranted, credentials or no credentials, just be sure that your point doesn’t depend on all of these people being crack dealers, because they weren’t.
And not everything on my post was about it, just to notice that you are falling for sources that are using the same tactics as the climate change deniers regarding what to do about science papers.
Like for example, nitpicking as if that had been my whole post. The argument continues to be about the weaknesses (that good researches do note and did), as if weakness is then just the only thing what we should consider. But that is not IMHO science.
As it turns out other researchers do still see it as science. Just not going down to the level of dismissing it because other researchers also reported that other studies show that this one is not happening a vacuum, there were other studies that show it can not be completely dismissed as the gun proponents want it too.
Even after all the critics try to make themselves so important (they can be, but their opinion only goes so far) one should be aware how important their opinion is taken by the ones that count:
So, yeah you are wrong here about your dig that I was just going about climate science, I was not talking about climate change, but about how papers in science are dealt with when they are indeed grossly wrong, It has happened before with anti-vaccine papers from Anthony Wakefield. The general point from critics has been that the paper is grossly wrong, but the point remains that if that was so then the same fate of Anthony Wakefield or Willie Soon would had befallen to Branas and many others.
It would be silly then to also just reply to me that “not everything is about vaccine science”. Hence the point here that you are doing that only to avoid dealing with one very important context: The critics are not making much headway in the scientific community. Like with other issues where science is not always agreeing with conservative politicians, right wing media or social media, their case then is made forcefully outside academia. It is not only a complaint here, but a warning to consumers of right wing media.
Again, case-control studies are one type of design, but not the only one. They certainly aren’t some kind of gold standard - they provide strengths and weaknesses.
You’re writing as if the adjustment made in a regression is proportional to the number of cases in particular categories. This is false.
Let me try to explain it to you. Imagine that you have a study with 100 men and 100 women, and you find out that being a man relative to being a woman explains 1% of the variance in cancer outcome. You’re writing about this issue as if a study that had 200 men and 100 women would find that being a man now explains 2% of the variance, or that a study with 400 men and 100 women would find that being a man explains 4% of the variance in cancer outcome. The amount to be adjusted in explaining cancer outcomes associated with gender is going to remain around 1% in each model.
You’re suggesting that I have to adjust 4 times as hard in the latter model than the former, but that is nonsensical. It’s a bit like suggesting that if I have a poll of 100 men and 100 women that says 51% of men favor Trump, a sample of 200 men and 100 women would find that 102% of men favor Trump.
It just doesn’t work that way. As long as you have sufficient numbers within a cell to get a decent representative estimate of the relationship for that cell, adding more people does not proportionally increase that estimate of the relationship. The variability within a predictor is a factor, and is taken into account in the regression modeling.
Sure, but what you’ve written thus far about location is not a functional criticism of the study. It’s coming from a place of ignorance about how regression modeling works. Move on if you like, but you’ve not explained how location is a failing of this study.
I agree with Hentor here but for context reasons, namely how seriously the critics in academia are considered, (kept in mind but not given too much sway) or how seriously word salads from the critics of the study that are not involved in related research are. (not seriously at all).
On edit, that was nice, but the reaserchers of 2013 point to other reserch that supported the maligned 2009 one.
I’m didn’t say it was your whole post. The rest I didn’t have comment on, so I didn’t. Like the rest of this post - there’s nothing of substance that is germane to the study at hand.
Then you’ve misinterpreted the criticism. It’s not a simplistic addition criticism as you describe.
Notice how you say as long as you have sufficient numbers within a cell? Do you think they did that perfectly? By your reasoning, in this study that used a little under 700 case example where 83% of them were outdoors, and just 9% of the controls were outdoors, you think the association they found would be unchanged if instead those percentages were swapped? And the method of conditional logistic regression would yield the same results? I’m not doubting the math that was used. But somehow having wildly different case vs control populations is just fine because that variation can simply be controlled for with regression. I’m saying it is not very convincing given the low levels of change necessary to render the results nonsignificant. Maybe I’m not communicating clearly given how far off your interpretation above is. Here is the same basic critique from link (about midway down the page) from DrDeth:
Moving on from location, the thing that drew me to this study was this part of the conclusion:
This isn’t well supported by both the data as well as the study methodology. The study doesn’t attempt to assess the protective value of gun use or possession at all - all of the case examples were taken from those who were first assaulted, then shot. It does not include any who avoided getting assaulted or those who avoided getting shot for any reason, including defensive gun use. How the study can conclude on the efficacy of defensive gun use without actually measuring that is a fatal flaw.
The idea of reverse causation is mentioned in the study, and is consistent with the results. Those that suspect they may be attacked may be more likely to carry weapons - those involved in illegal activity, etc. And when we look at the breakdown of the case vs controls, they were more involved with illicit drugs, in areas where there were higher illicit drug trafficking arrests, and higher history of prior arrests.
The controls were also taken from the city as a whole, not accounting for different areas of a large city have different levels of crime, etc. Nor did the study take into account any firearm training, length of ownership, etc. It treated a person as possessing a firearm for the control group “if they reported any guns in a holster they were wearing, in a pocket or waistband, in a nearby vehicle, or in another place, quickly available and ready to fire at the time of their matched case’s shooting.” But of course, for many gun owners if some rando calls them on the phone and asks them about their guns, they aren’t going to be forthcoming.
All of these are weaknesses in the conclusion drawn from this study.
/cdn.vox-cdn.com/uploads/chorus_asset/file/9632497/swanson_slide2.png){kind=link}