Sure, the evidence seems to nullify the nonsense that the NRA has been spouting for decades… But they’ve been impervious to the facts for all this time, I can’t see how this would change any of their minds. (Boy Howdy I’d love to be wrong about this!)
1 Like
E pur si muove.
7 Likes
Going just by the numbers in the letter:
161 injuries expected, -
129 injuries observed =
32 decrease in injuries vs /
12.7 ( sqrt(161) ) standard deviation =
2.5 sigma claimed effect
And that doesn’t take into account any of the P-hacking they may have done.
Standards differ from field to field, but personally I don’t get out of bed for anything less than 6 sigma.
(Literally. I am a scientist who gets a page and text message at any time of the day or night when certain events occur with >6 sigma significance. I then confer with other team members and in the next ~15 minutes we write up a report to other scientists describing the event. These reports sometimes include a phrase like ‘Due to the low significance of this event (6.5 sigma) […and other reasons…] we believe that this is a statistical fluctuation and not a real event.’)
Remember guns don’t kill people…actually guns do kill people, but that’s not the point…the point is that bad statistics also kills people.
1 Like
I believe this is modeled after a 2015 study that famously found lower mortality for high-risk heart patients admitted during a major annual cardiology conference. Fewer risky procedures being done, apparently, which you might expect to have a very brief displacement effect, but deaths were still lower 30 days out so it implies the risks aren’t always worth it. https://www.ncbi.nlm.nih.gov/m/pubmed/25531231/
2 Likes
I think another important thing about this study is that it exists at all. With the NRA’s push to prevent the CDC, and anyone else, from studying and accumulating data related to gun injuries, to get this info out into the open is refreshing.
It’s instructive to note that though most things conservative leaning people don’t like or disagree with tend to require more study and data, analyzing things they do agree with should be banned.
How do the adherents convince themselves they’re not completely full of shit?
4 Likes
Could this be what’s going on? Places being closed because the owner is at the convention?
I have a big problem with the data as presented–the effect is simply too large. If it really were due to those 80,000 they would have been sued into bankruptcy long ago from all the injuries they were causing.
1 Like
Yet there are commenters up thread who are skeptical because they think the measured effect it too large.
On the other side of the coin, this effect was observed in a situation where the convention-goers were without their guns for only 8 to 10 hours a day. I think it’s safe to assume the majority of them brought at least one gun with them to the location, left in their vehicle or hotel room (or they were within day-trip distance and went home every night). So this effect wasn’t just observed with complete absence of guns for a week, but only absence of guns for a partial day for a partial week.
You’re a statistician and your accusing people you don’t even know of p-hacking? P was < 0.004. P-hacking usually only occurs when the value is close to but shy of significance. The measured p-value was more than an order of magnitude on the side of significance.
at the gun industry or any shooting range getting sued (successfully, that is). You did notice that this study took place in the US, right?
3 Likes
An effect can be simultaneously too large to be credible and too small to be statistically compelling: Suppose I notice two car accidents with even numbered license plates and one with odd. The 2 vs 1 is not statistically significant, and it is not credible to believe that odd-numbered cars are twice as safe as even.
Likewise I don’t think it is a priori credible that <1% of the gun-owning population attending a convention can cause a 20% decline in shootings, and it will require a lot more than 2.5 sigma to convince me otherwise.
dmpalmer:
And that doesn’t take into account any of the P-hacking they may have done.
Just look at the garden of forking paths:
We can hypothesize:
- Higher/lower rates
- compared to the population/total medical claims/emergency room visits/deaths
- of accidental/intentional/all
- killings/hospital admissions/ER visits/arrests/accidents/crimes
- with/without/regardless of insurance
- involving/used by/used on/mentioned guns
- overall/in gender/in age range/in region/in state/race
- within X days of
- NRA/Republican/Democratic/ACLU
- state/regional/national convention
- All/militaristic/family/secular holidays
- sports events/specific sports events
- {all, rap, country, rock} music festivals/concerts
- NRA/Republican/Democratic/ACLU
Admittedly, these hypotheses grind different axes, and the study wasn’t sensitive to all of these branches. But at only 2.5 sigma, you really need to pre-register your hypothesis before you do the statistics.
Taken at face value we have 80,000 people responsible for about 3,000 injuries/year. That would mean most NRA members accidentally shoot someone.
That doesn’t make sense.
If the effect isn’t statistical noise it’s the absence of those people causes other changes.
1 Like
This was published in a medical journal, where the standard for comparing two groups in a sufficiently large sample is the student t-test or a chi-square test, with a standard threshold for statistical significance of p<= 0.05. Keep your six sigma to process control, please. If the medical scientific community required 6 sigmas for significance, we’d still be pondering the relationship between clean water and cholera.
2 Likes
So your argument is simply:
It isn’t. They did extensive multivariate analyses to determine if there were other factors behind the numbers. There weren’t.
Which is a major part of the conclusion the authors reached.
5 Likes
Eh - that really hasn’t stopped outrage before. 90% of the shit that comes out of his mouth is just words that mean nothing. It’s just another example of his authoritarian nature AND his ignorance of how the fucking law works.
Suspending due processes for anything is a horrible idea. I can’t imagine there NOT being out rage at the suggestion that we “take away or limit something, and then go back and get court approval”. Even if it’s just him spewing the same nonsense as normal. It’s dangerous nothing because, like all the other nothing, it makes it seem like a normal thing. Or it makes shit like insinuating the death penalty for drug dealers seems reasonable by comparison.
ETA - woah- I dunno if the preview is YUGE for anyone else - but I didn’t do anything to make it like that.
3 Likes
If the medical community were less willing to accept P=0.05 results, then we wouldn’t have the case where one in twenty ineffectual (sometimes dangerous) quack cures get the seal of approval.
See Ioannidis, 2005, “Why Most Published Research Findings Are False”
We are left to decide between two propositions:
- The results of the paper cannot be taken at face value
- Each NRA convention-goer causes a shooting every five years, which he could avoid by going to an NRA meeting every weekend. (This counts only the shootings of insured people on weekends: so multiply by a factor of 5 or so for the total damage)
Just because a paper is published, doesn’t mean that you have to believe it.
Nonsense. If statistical significance for clinical studies were as stringent as you think they should, no progress would ever be made. In your (imaginary) world, there would still be controversy about whether surgeons should wash their hands, FFS! The NEJM would publish their annual (and only) issue and it would have “THIS PAGE LEFT INTENTIONALLY BLANK” by page 8.
3 Likes
As all of them apparently are…
2 Likes
I actually agree with you that Trump’s comment about “take guns first - due process after” was some whacked out authoritarian shit. But I want to philosophically quibble about the balance of individual rights against the power of the state.
There are already processes in place where a person can basically make an administrative decisions to trample all over someone’s rights. They kick in generally during crisis situations. Like a psychiatrist can have you temporarily locked up basically on their say-so. Children’s Aid workers can take your kids away. In a crisis situation police can restrict a lot of your freedoms.
So imagine if there was a system where a when a psychiatrist had you involuntarily committed, the cops also came to your house and took away every gun they could find, and you lost your right to own guns, and had to do something to have that right restored after your involuntary commitment ended. You’ve already been identified as a person so dangerous to yourself or others that you had to have your freedom to walk down the street and make your own decisions taken away, at that point going after second amendment rights as well doesn’t seem outlandish
I bring that up because sometimes I feel like discussions of rights get inverted. Locking someone up or taking their kids are both much more extreme than taking away guns. I think that Trump’s comment might have struck some people as being somewhat reasonable without those people being crazy authoritarians. My suggestion of taking guns when you are involuntarily committed might be a terrible one (you know, since it was totally off-the-cuff) but unless someone was generally against having people involuntarily committed it at least sounds like a not unreasonable thing.
Of course that’s not what Trump meant. He meant some crazy police state nonsense because he lives in a cartoon fiction where there are good guys and bad guys and so you just have to good guys take the guns away from the bad guys. Just like when he talks favourably about the death penalty for drugs he’s actually praising Duterte for forming death squads that undoubtedly killed tons of innocent people. Just like him musing about how maybe America should have a president for life.
3 Likes
Yeah, but in this case, it was clear he didn’t mean it from the start, given the support he’s gotten from - and vocally given - the NRA. As usual he didn’t know what he was talking about and fell back on authoritarian blather, but he didn’t mean to say something that would upset the NRA. Once he realized he did that, he completely reversed positions to the other extreme (of not allowing any changes to happen at all). Even during the meeting itself, it echoed his meeting with legislators about DACA, where he urged them to pass something and promised to sign it (and then privately refused to sign anything they passed); given his history and NRA support, it was clear from the start that the same was going to be true in this case as well.
Executing drug dealers is something he’s voiced support for a long time, on the other hand.
Well, sometimes the words mean something to him.
1 Like
This topic was automatically closed after 5 days. New replies are no longer allowed.
