In fact, I do know better. This is what happens when one follows several similar threads featuring the same members in fora with different rules and gets careless. I apologize for violating GD rules because I carelessly replied thinking it was the “stupid gun news” thread running in The Pit. No excuses.
The reason we engage him is because he keeps insisting that science is on his side in much the same manner that Pat Roberts keeps insisting that God is on his side. I’m sure he believes it but he is just wrong.
Apart from all of the peer reviewed publications in scientific journals I keep citing, of course. Somehow, each and every one of those can be handwaved away (because they reach the wrong conclusion).
When you have to explain how all the science is wrong but it gets published anyway due to a conspiracy, you’re a conspiracy theorist, not a scientific thinker.
You mean all those studies published by people on the gun control side that don’t say what you think they say but cleverly invite you to reach the conclusions you have reached?
When I point out that the study does not state that the presence of guns is the cause of the increased incidence of murders in households that have guns; that other things like being single or renting your home or being a habitual drug user are far more correlated to being murdered than someone in your household owning a gun; that (as abates points out) gun ownership seems to do very little to increase the likelihood of being murdered among the cohort of people who have other risk factors (IOW maybe the people who are getting murdered have guns because they are at high risk of getting murdered rather than being murdered because they have guns, i.e. the causality is reversed). You just continue to act like the study is actually saying that bringing a gun into your home will increase your chances of getting murdered several times over.
I agree that the studies are constructed to invite the reader to believe that some suburban chubby white dude that buys a gun and brings it into his house has suddenly doubled the likelihood of being murdered but it doesn’t really actually say that the chubby white dude is now twice as likely to be murdered now does it?
>>>>>>>>>>>>>>>
Or do you mean all those pro gun rights studies that YOU handwave away because they reach the wrong conclusions?
When I invite you to present a study, ANY study on defensive gun use that comes up with a number lower than the NCVS and you reply by attacking the saying that the range of existing estimates is so wide that you can’t rely on any of them (so we should act like the number is effectively zero). Hemenway criticizes the most extreme estimates of defensive gun use but he does not seem to claim that the number is negligible.
>>>>>>>>>>>>>>>>>>>>
Hemenway does not say what you think he does. While I suspect that he would like to ban guns entirely he argues for licensing and registration rather outright gun bans because gun bans because gun bans are really hard to defend without looking retarded.
Again, these are peer reviewed scientific publications. Your description of this as a “side” and as “invit[ing] you to reach conclusions” means that you think the editors of these journals and the reviewers of these articles are conspiring with the authors to publish unscientific papers. You are a conspiracy theorist.
When I point out that the study does not state that the presence of guns is the cause of the increased incidence of murders in households that have guns; that other things like being single or renting your home or being a habitual drug user are far more correlated to being murdered than someone in your household owning a gun;
[/quote]
You don’t seem to realize what you’re saying here. If there are other covariates in a model that are “far more correlated” to use your odd word choice, AND gun ownership is still significantly associated with homicide, it means that despite accounting for the covariate, gun ownership still conveys significant risk. Think hard about that.
What that means is that the factors controlled for cannot explain the association with guns and poor outcomes. Otherwise, it would be reduced to nonsignificance. The significant covariates (and matching procedures) undermine the reverse causation argument, because they account for the factors you want to claim are still relevant. If you could understand the science, you would comprehend this.
But you just got done observing that other covariates ARE in the model, and gun ownership STILL conveys double the risk. Or, in other models, individuals are matched on these risk factors.
You want to pretend that in these studies, the one group consisted of people who brought death on themselves through their different behavior and lifestyle, which led them to both get a gun and be killed by yet some other gun. That position is not consistent with the matching procedures and variables that are controlled for, which are designed to equalize across those different behaviors and lifestyle. It’s also not consistent with the fact, as I observed to you before, that the most frequent circumstances of death are arguments among family members.
Studies, plural, and meta-analysis, and they do indicate that having a gun in the house is associated with double the risk, over and above whatever covariates they have been able to measure.
If you want to be taken seriously, stop being an antiscience conspiracy theorist. I’ll be happy to try to keep helping you understand them, like explaining what a covariate is and what it means to have it in the model (as I did above).
No, the articles don’t say what you think they say, they just invite you to reach those conclusions. Why would this require conspiracy? The articles say things that are completely true and create a path that leads many people to conclude that bringing a gun into your home more than doubles your chances of getting murdered. The article invites you to conclude that if a chubby white suburban dude brings a gun into his home then everyone in that home has just doubled their chances of getting murdered. But the article never says that, in fact it says that you can’t infer this, but you still do it anyways. I suspect that they knew you and others who are inclined to see guns in a negative light would do this.
I think you don’t understand what I am saying. If you spent a little more time trying to understand the arguments and a little less time trying to convince everyone that you have some special credibility because you are a scientist, perhaps you would understand what I am saying better.
Those other factors (like being single, or living alone, or being a habitual drug user) do not represent the universe of things that might increase the likelihood that someone will get a gun because they think they are more likely to need one.
Sure, about 25% of ALL murders are committed by family members. The fact that about the same percentage of murders committed in a household with a firearm are committed by family members doesn’t really seem that surprising. You conclude there was a murder because there was a gun in the house, perhaps you should conclude that there was a gun in the house because there was an elevated chance of murder.
Yeah, there is a correlation. I don’t argue that. You are trying to twist that correlation into come sort of causation. You are trying to make the studies say something they don’t say. Something they CAN’T say. And as a scientist, youa re comfortable with that.
I’m not anti-science any more than than the next guy. I just don’t think the study says what you are trying so desperately to convince everyone the studies say.
Can you explain to me why a single person didn’t have an elevated risk of being murdered by having a gun in the house? Doesn’t that mean the “double risk” is a generalized conclusion? I don’t really understand how the study accounted for this fact. Did it just average these folks in? How does that work?
Can you also explain to me again how bringing a gun in the house causes the elevated risk of murder? The article seems to be pretty clear that the murders weren’t necessarily committed by the gun that was brought into the house.
You are damn near asking for certainty before you will acknowledge that there are a significant number of defensive gun uses and at the same time you seem to be willing to wave away the caveats that are in the studies themselves in order to reach a conclusion that the studies don’t actually reach, they only imply.
Sure, there are some increased risks when you have a gun in the home. Gun accidents are much likely with a gun in the home than if there isn’t one, but I think the numbers would be a little underwhelming relative to the hysteria that guns produce. Sure there are some arguments that would have led to severe beatings that would be replaced by someone getting shot. But the articles don’t even come close to saying that having a gun in the home causes the increased likelihood of murders. And you want to pretend that guns are to murders what cigarettes are to lung cancer, after all, its all just correlation right?
The difference is that I don’t think that people who are more likely to get lung cancer are going to be more likely to go out and start smoking and the causation might be reversed (increased likelihood of cancer does not cause an increased likelihood of smoking). I suspect that someone who thinks they are likely to get murdered might go out and buy a gun.
From another family member? So, they got the gun as protection from the family member, but then the family member got some other gun and shot them with that?
It is an association. So you test hypotheses to rule out alternative explanations. You include other potential explanatory covariates, and they make those other explanations more or less plausible. And so far, accounting for the behaviors and lifestyle factors that might put someone in fear of their lives doesn’t explain the increased risk of having a gun in the house.
By the way, since you brought up cancer and smoking, do you believe smoking causes cancer? Please point to the study that proves it? AS I’VE TRIED TO EXPLAIN TO YOU REPEATEDLY, NO STUDY PROVES CAUSE. You’re handwaving from an antiscience position of ignorance.
You really don’t grasp how this undercuts your alternative explanation do you? That’s kind of awesome. Think hard about this: If your reverse cause explanation were true, people buy guns because they think they are at risk. Thus, there should be no difference in risk associated with having a gun if you live alone or with others. On the other hand, if reverse cause is not true, and it is the presence of the gun that leads to the increased risk of homicide to household residents, THAT REQUIRES HOUSEHOLD RESIDENTS to be present. Otherwise, it is a suicide!
Wait, what? Let’s back up. Do you know what a regression analysis is?
Which article are you talking about? There are many.
No, the point is that you see this huge causal effect and point to the ~30% of murders committed by family members when there is a gun in the house and I am pointing out that ~25% of ALL murders are committed by family members regardless of whether or not there is a gun in the house, I don’t know enough to say whether this is within the standard of error or if this is enough to explain the more than doubling of the murder rate but it is not evident on its face that this is the case.
I understand they try to compare apples to apples (or is it something more than that; but when they look at the group of people who are living alone (for example), do the people who live alone have double the likelihood of being murdered if there is a gun in the house? If not then why not? Is it possible that there are other things that increase the risk of being murdered and that living alone is an indicator of those thing but not a direct cause of the increase in risk, perhaps owning a gun might be an indicator of the presence of things that increase risk without actually being a significant contributor to the increase in risk. Or are you certain that these studies accounted for all possible variables?
So you ARE saying that this study provides the same sort of conclusivity that studies on cancer prove?
So I can say that the presence of guns in the home might be an indicator of other things (that are not accounted for) that might increase the risk of being murdered. In what way could smoking be an indicator of other things that are increasing the risk of cancer that are not otherwise accounted for?
So you are basically saying that virtually all of the elevated is the result of people being murdered by other people that are living with them? I think you are once again saying things that the articles don’t say.
Yes, I think its how you determine how one variable affects outcomes when you have a lot of variables to deal with. I know that there are a lot of different ways to do it but it but it doesn’t determine causality or even the direction of causality if there is any, does it? It is certainly a useful predictive tool about the coincidence of two things in the current population and sometimes one is causing the other and sometimes both are markers for the same thing. But it is not necessarily a good predictive tool about what would happen if you artificially increased one of the factors. That would be a better indicator of causality, right?
If I take a random population of people (and correct for all the things we would normally correct for) and forced them to start smoking two packs a day for 20 years, I am willing to bet that we would see an increase in cancer rates.
OTOH, ice cream sales are correlated to increased murder rates. If I dropped the price of ice cream to $1/gallon and increased ice cream sales, I doubt it would have any effect on the murder rate.
Similarly, if we forced some random selection of people to keep a gun in their house, I doubt that the murder rate in those homes would more than double.
All this to say, regression analysis and the other stuff you seem to think carves the “guns in the home cause higher murder rates in the household” in stone is really not carved in stone and there could be a lot of other reasons why those people are getting murdered at twice the rate of people who do not have guns in the home that does not come from a causal link between the presence of guns and murder. You keep saying “SCIENCE!” as if science is telling us for a fact that guns are causing the increase in murder rate, when it does no such thing. Or does it?
ISTM that you are abusing science by claiming that these studies give us some conclusive level of evidence that guns in the home are the cause of the significant increase in murders in those homes. So maybe you don’t have as much respect for science as you claim.
I was talking about the original one. I am aware that they are many but don’t most of them just confirm what the first one says? That the presence of a gun in the home is correlated with a significant increase in murders in the home? I don’t think anyone is denying that this correlation exists, the fact that there are multiple studies that say this doesn’t really help you with the notion that there is some sort of causal link does it?
You know, there’s just so much wrong with this post, and there’s so much of it! I’ve been loathe to come back to it because it feels like so much work. Could you aim in the future to be a little more concise? Try to focus a little bit more because sometimes you’re all over the place.
I decided I would take this post by addressing the conceptual threads and I would start with the ignorance of science that is represented by the above. You keep trying to mock me for being a scientist and then you generate this kind of thing!
Replication may be the number one most important aspect of science. It isn’t that there’s one study with a finding. All of these things that you keep pointing out – “correlation is not causation,” “this study has limitations,” “this study doesn’t control for this factor”… all of those things are true about EVERY study. That’s why replication is the key. That’s why reliability is so fundamental (i.e., not getting dramatically different estimates across studies). When different studies, especially from different research labs, produce consistent findings, that helps to increase the confidence that things like the limitations of a study, or the different variables controlled in a study, or the unreliability inherent in a measure were not good explanations for the results.
We believe that cigarettes cause cancer not because one study found that. NO STUDIES PROVE CAUSATION. It is because of replication. Replication in different samples, different populations, different measures, different control variables. Alternative explanations become less and less plausible.
So far, the studies on the increased risk of homicide associated with having a gun in the home is found consistently across studies. The replications are helping to make alternative explanations less plausible. More studies with better measures will be very helpful, which is why the NRA does not want those studies to be done. But they will be.
A couple of things though: how many of the independent studies might have the exact same bias, simply noting the presence or absence of guns without any deeper analysis, being motivated by seeking the same conclusion? How many independent studies concluded that African-Americans were innately less intelligent than whites for example? To treat guns as vectors of a deadly disease (death by gunshot) is hopelessly simplistic, yet it almost seems as if that’s the goal of the research, to declare guns=death, case closed.
Think about it a moment: how could owning a gun cause violence and murder to take place where it otherwise wouldn’t have? The only plausible causal mechanism would be for guns to somehow inspire their own use. That was the argument originally made against Shall Issue gun carry, that carrying a gun would release peoples’ murderous id and cause a skyrocketing gun homicide rate. But that simply wasn’t the case. Much has been made of opposition to doctors asking or reporting the presence of guns in their patient’s homes, but look at it from the gun-owning patient’s view: A stable middle-class home occupied by people with no history of violence, drug or alcohol abuse, or criminal activity. Yet mention that there happens to be a gun in the house and suddenly the perceived risk of danger is quadrupled. Attempts to look at gun violence statistically are imperfect because people are not interchangeable units like molecules of gas. The risks associated with owning a gun are probably somewhere in the slippery middle between purely mechanistic and purely statistical.
Think about it for a moment: how could exposing a lung cell to cigarette smoke cause it to turn into a cancer cell?
Maybe there are some good answers today. But I don’t think there were good answers when the decisive studies proved the link between lung cancer and smoking.
Just to draw out the implication of my last post: We don’t have to prove the mechanism to draw conclusions, if the evidence is strong enough that the relationship isn’t coincidental. Consider #6 here:
There does need to be a biological plausibility. That’s a low barrier. Surely it is biologically plausible that having a highly effective tool for a job makes it more likely the job will be done.
It could be that guns kill more because physicians are worse at treating gunshot wounds than those caused by knives. Or it could be that if is often less frightening for a perpetrator to pull a trigger than to stab someone. Or it could be that having the gun tool, through some psychological mechanism not now well known, increases anger. Or it could be that perpetrators shoot back when someone pulls a gun on them. Or it could be something else. We don’t need to know the biochemical pathways to document a relationship.
I refer you to the Spurious Correlations site, which proved that eating margarine causes divorce in Maine.
Where did they prove that? Beyond which, what is your point? To repeat myself, observing a.correlation is where science begins. Didn’t you see what I’ve written above? That’s when you start systematically testing hypotheses. That’s when you start attempting to control for potential explanatory factors. As is the work being done regarding guns and homicides in the home.
Or are you alternatively arguing that all correlations are spurious? That would be a weird way to go through life.
You all seem to see “correlation is not causation” as some kind of discussion ending trump card. Yet you also seem to believe that cause still exists in the world. Doesn’t that raise any dissonant sensations in your head?
I was responding to stuff that you posted, so if my posts are all over the place it is because I am responding to your post.
You make a big deal out of how many of these murders (in households with guns) are committed by family members or other members of the household; ~30% of all the murders committed in homes with guns are committed by family members. And at first glance, that might make a good case for a causal relationship between guns in then household and murder. After all family arguments that would have resulted in a battered wife might now result in a dead wife (or less frequently a dead wifebeater). But ~25% of all murders are committed by family members regardless of whether or not there is a gun in the home. Like I said, I don’t know if this is enough to support the notion that you are more than twice as likely to be murdered if you have a gun in the house? Does it even get us most of the way there?
I am perfectly willing to acknowledge that there might very well be some murders that would otherwise have been trips to the emergency room, etc. if there wasn’t a gun in the house but its not quite as simple as saying guns in the home more than double your risk of being murdered. Is it?
You basically reach all sorts of conclusions that are not supported by the study. You see people who live alone having no elevated risk of being murdered due to the presence of guns and somehow conclude that this means that the elevated risk is largely due to people murdering members of their household, especially when people who live alone have a much higher chance of getting murdered than the general population of people with guns in the household? How does it not occur to you… a scientist… that perhaps the same thing that is causing the elevated risk of murder in households with guns is also associated with people living alone (or renting or doing drugs, all things that increase the chances of being murdered).
You asked me about regression analysis and I gave you my understanding of how regression analysis is used to control for variables. Is there something wrong with my layman’s understanding of the concept of controlling for variables?
I’m not mocking you for being a scientist, I’m mocking you for acting like it gives you some extra layer of credibility that relieves you of the requirement of making your case. Instead you just say “jeez, I can’t explain it to you laymen but these studies say what I claim they are saying, and its as simple as that.”
And when all the studies don’t control for the same possible explanations, then it doesn’t really mean very much that they all find the same correlations, does it?
Wait. What? So if I keep doing studies that ignores or can’t account for some possible explanation, the fact that all these other studies that all ignore the same possible explanations all came up with the same results somehow means that the ignored possible explanation somehow becomes less and less plausible?
Like I said, I am willing to bet that if you took a random population of people and forced them to smoke two packs a day for 30 years, you would get a significantly increased incidence of cancer.
Would you be willing to bet that if we gave a random population of people a gun and told them to keep it in their house, that would result in the murder rate of the people in those homes to increase by anything approaching 240%? Do you really think we are at the same place with guns=murder and cigarettes=cancer?
I generally think that the more we know, the better off we are. I think politically driven junk science eventually proves itself to be politically driven junk science. The legislation prohibits the agency from promoting gun control and I think that if gun control is where the facts lead us then so be it. I would repeal the law if it were up to me.
I am perfectly willing to admit that there is a high correlation between guns in the home and being murdered. But when you go around saying “gun owners are twice as likely to be murdered. It is that simple” You are glossing over a whole lot of shit.
I think it can make a difference at the margins. A wifebeater that might otherwise send his wife to the hospital might kill her if there is a gun available. The study does not seem to differentiate between households that included prohibited persons and households that did not. The studies do not seem to control for gang membership or prior violence in the household. The studies don’t control for a whole lot of shit that we KNOW increase the chances of murder. Controlling for race and socioeconomic class while not controlling for factors like prior criminal or arrest records seems a bit counterintuitive.
Really? Because you seem to be saying that this is where the science ends, we are pretty much done. We’ve proved correlation and we can never prove causation so the link between guns in the home and murder is like the link between smoking and cancer and anyone that argues with that conclusion is anti-science.
And yet all the studies seem to account for the same variables without accounting for prior arrest records and criminal history of people in the household.
Its not that. Its that you see correlation and are very eager to leap to causation (because of your views). We see correlation and we are skeptical because we see other things that might be the common cause of gun ownership and getting murdered (because of our views). Things that criminologists have long associated with murder, like gang association, prior history of violence, criminal record or arrest record. I mean, how do you NOT control for these things and claim to have done anything approaching a thorough job.
I was responding to PhillyGuy when he said
Look; for the record I acknowledge that of course more gun deaths are going to take place in homes where guns are present than in homes where they aren’t; that’s common sense, on a par with saying that people who buy fireworks are more likely to sustain fireworks injuries and deaths. What I’m disputing is that absent all other factors, a gun will in and of itself cause a significant increase in gun injuries and deaths. Factor out high-risk groups like felons, cases where a gun was premeditatively obtained for the purpose of committing homicide, and situations where ownership of a gun was coincidental, immaterial to the circumstances of death, and I strongly suspect the actual risk of owning the gun itself is far smaller.