The claim isn’t as strong as you state. Buybacks occurred after the study date, but they also occurred during the study period. In addition, there were overlapping amnesty periods that firearms could be turned in. In other words, inconclusive. Not a fatal flaw. It shows some results, but it’s not a great study since it doesn’t account for the long tail. I can’t seem to find any study that focuses on substitution over a longer period.
I would say that examining within country differences over time is valuable - to those countries. That value ends at the border. That’s really my point about the Australia information. While it’s interesting by itself, it doesn’t inform discussion as to the US.
Plus, every time I hear the phrase, “Gun owners need to compromise,” I just about have a shit hemmorrage.
“Compromise,” in the context of the larger gun rights/gun control debate, has become a meaningless word, and I now refuse to even acknowledge the existence of anyone who brings it up.
And yet the both of you have absolutely no problem participating in these threads, if only to spread venom and distrust. You call anyone that doesn’t fully support your point of view on this subject a dishonest extremist-are you just here to call people names?
So there is no way to determine if “A” might causing “B” or whether “A” is almost certainly causing “B”? Is there a way to determine whether “A” is causing “B” rather than the other way around?
*I’m not a physicist so this is heavily abbreviated and simplified.
Essentially, empirical studies cannot prove causation. Like, ever I think. They can show correlation, in different forms, at different degrees, and with different variables. Essentially if you add up all the different types of correlation, correcting for all the crap that happens in real life, you can theoretically paint a pretty good picture. And by pretty good, I mean like gravity.
So the challenge for any data set that shows these varying levels of correlation is to demonstrate how well or poor they adjusted for these different interfering real life crap, how well the data shows the correlation that is being implied, and to determine if contrary scenario are available.
So rather than saying that a certain study doesn’t prove causation which is a weak argument, it is a stronger position to challenge the study methodology if it can be shown to be flawed. The 2.7 times figure by Kellerman for example, whose sample population was not representative. Since extrapolation on non-representative samples would lead to inaccurate results, if you can demonstrate that the sample was not representative and simultaneously not adjusted for, then it is a strong argument against the conclusion.
The study here in Australia seems to indicate that a massively confiscatory law did in fact correlate with a reduction of firearm homicide and suicide. The caveats are that they were dealing with small numbers to begin with, there were simultaneous efforts at public awareness and reductions ongoing, there were downward trends in certain subsets of these populations, and there may have been significant substitution among certain demographics. And…Australia is sufficiently different than the US that any conclusion that can be drawn for that population would not be applicable to a US population.
I haven’t followed this thread for a while. I just read the above posting, and from it I couldn’t tell which side Scumpup is on. Word for word,* it eloquently expresses my exact experience –as a gun control advocate.
*Okay, maybe I’d be less inclined to use the term “militant” to describe myself.
If you have some complaint, report my post. I have nothing to say to you and there is nothing you have to say that interests me in the slightest, least of all your junior mod scolding.
What good does it do to hop into a thread just to tell people that you have no interest in discussing the topic because you don’t trust anyone with an opinion that differs from yours?
The study ends in fucking 1998. It doesn’t even account for a short tail! It has no tail!
There are studies that look at over a decade after the gun legislation. I think you mean that you just cannot find one that concludes what you want it to conclude.
Really? Australians are just categorically different than Americans? So much so that their experience with firearms has no relationship to our experience with firearms? That’s just stupid. We interpret data and analyses from all over the world for all kinds of purposes - medical studies of French people, economics studies of England and Japan, criminological studies of the Netherlands…
Why is it that when it comes to guns, there’s no comparative information to be learned? What’s special about behaviors related to guns that makes it so categorically different?
The answer is that there isn’t any such magic specialness to guns. This is just more categorical thinking from gun nuts.
The main product of regression analyses is a regression line. For example, imagine an equation like this: y = a + b1x + b2x2…bnxn + E.
This equation is derived from a set of data points with the intention of making predictions when you plug in new values. You could imagine that y is “violence score” on some continuous measure of violence, and that x1 is “gender,” and let’s just leave it at two values for gender in this data set. They could be coded as 1 for boys and 2 for girls.
So, the beta weight - b1 tells you how much the predicted value (y) changes for boys relative to girls. If the weight is 5.7, say, that means that the value of the violence score y goes up by 5.7 for boys compared to girls.
Let’s say that b2 is IQ, and x2 is -0.5. That would mean that for each increasing unit of IQ, the value of the violence score y is reduced by 0.5.
What I’ve quoted above makes it clear that you are not interested in learning, you’re interested in some competition where if you can identify a limitation to a study, you’ve undone the study altogether. You sound as if observing that there are limits to the generalizability of a sample, you can just brush your hands and walk away saying “My work here is done.”
How many studies do you think have no limitations whatsoever?
We have to look at individual studies, understand their limitations, and take the results for what they are worth GIVEN the limitations.
Given the limitations of individual studies, though, replication is the key. Replication is how a body of knowledge accumulates. So, while there may or may not have been problems of generalizability associated with the Kellermann (1993) study, when you see a similar estimate of risk in the Dahlberg et al (2004) study, you feel more confident that the conclusions of Kellermann were not undone by the generalizability issue.
When you see a meta-analysis that combines data across studies, like Anglemyer et al (2014) that finds that the risk for death by homicide is about doubled when a gun is in the home, you feel very comfortable that the Kellermann sample is not a fundamental flaw.
Well, I appreciate your contribution, as it represents a nice example of the content-free kind of post I’ve come to expect from gun advocates. I point to it as evidence in predictions for other people about the capacity for thoughtfulness and rationality that is possible from gun advocates.
Gun buybacks began in 1996. Short tail, no tail, however you’d like to characterize it. It does include a portion of the period of the law in question. Your previous statement that it had a fatal flaw implying it didn’t measure the results period has merit, but not as strongly as you suggest.
I said I could not find one - at all. You seem to have access to scholarly research type places that typically require registration. Can you find one? The type in question is one that has a longer study period, and focuses on substitution or at a minimum, draws conclusions on substitution.
Yes, Australians are categorically different than Americans when it comes to guns, history, culture, laws, etc. The degree of that difference and the impact it has on comparisons and their relative value is surely debatable. And while I think in total there may be some comparative value, I find it so small as to be meaningless. So not that it has no relationship, but none worth expending effort over. You expressed similar thoughts, though the degree is different:
So the question is really the limitations. That is somewhat of a tertiary topic. First you’d have to conclude about the US, then about the foreign country, then about how that foreign country relates back to the US, with each step along the way introducing noise. What parallels do you think you can draw then, and how do you control for the differences? In other words, to what degree is the comparison limited and how would you mitigate those limitations? That’s what you have to answer when you introduce foreign country experience.
If restrictions on firearms did not result in an overall reduction in harmful outcomes (e.g., homicides, suicides, injuries, accidents and gun crimes), then I would agree that the restrictions were not worth it.
I wasn’t aware that refuting evidence is an indication of lack of interest in learning. This is a debate forum after all, perhaps you’ve lost your way? Can you identify a factual error in the part you quoted from me?
In any event this is essentially what you’ve done [walk away] when you dismiss the results of the NCVS survey because the data is self reported. I’m just not drawing a conclusion about your interest in learning.
Depending on the nature and magnitude of the limitation, yes there are such that would be, as you say, fatal flaws. Not to say that the Australian study in question suffers from this. The study itself identified the same weaknesses I did (it’s where I pulled them from after all), and I do tend to agree with their conclusions. Magnitude of substitution, the predictive value and how those would translate to other countries was not part of the study and are really the main areas of my interest.
Without commenting on the value of any of the above referenced studies (if you wish to discuss those, please present their findings with a link to the actual studies), it’s great that you understand the value of replication. So inearlier threadswhere we talk about DGU, and there have been at least 13 other studies that have all estimatedmultiple hundreds of thousands of incidents, you wouldn’t just disregard them all since they are all directionally consistent. Oh wait, that’s what you did. Your stated reason was that you can’t rely on self reported incidents - but of course the Anglemyer study you reference above uses firearm accessibility by survey interviews mostly.
And the Dahlberg study suffers from the same weaknesses as other studies that purport to show similar results - that it can not be inferred whether people are acquiring firearms as a result of their already existing environment, or the firearm acquisition causes the circumstances in their environment. Not only that, like other studies in this fashion, it doesn’t distinguish between events where the firearm present in the residence are used or if the firearm used was brought from outside. That is a critical piece of information to determine any kind of causal theory.
We were having this same discussion in the other thread, and then you decided not to respond any further. Feel free to respond topost #63 in that thread.
Great, thanks for answering. So you acknowledge there is a cost / benefit aspect to how we should craft firearm laws. You have been diligently pointing out the costs, how have you analyzed the benefits to come to your conclusions?
What are the 13 studies? How is 65,000 consistent with 2.5 million? That’s a hell of a lot of rounding there.
And yes, they all suffer from the same fundamental flaw - a flaw that is revealed when one actually looks at the reports. Show me the results from surveys that remove the incidents of self-defense started by the informant, or that are obviously fabricated.
Of course if you see 65,000 and 2.5 million as consistent, your judgment might be too far gone.
Who is saying that? Perhaps your judgement is too far gone because you are seeing things that aren’t there? Differences in estimates are expected due to methodology differences. Regardless, all point to a significant number of DGU. But it’s great that you can rely on data that is replicated across studies to form conclusions, just only in areas you agree with, right?
Any response to this:
But hey, at least here you’ve mostly not omitted relevant information or changed the wording of various conclusions from the studies you are referencing in this thread. Congratulations on that.