On a legal academic discussion list, someone said — quoting the Brady Campaign that “In two thirds of battered women’s households that contained a firearm, the intimate partner used the gun against the woman, usually threatening to shoot/kill her (71.4 percent) or to shoot at her (5.1 percent).” Given that there are a lot of women who are battered by intimates — the National Crime Victimization Survey reported 900,000 in 1998 — that sounds like a vast number of death threats.
But if you actually look at the cited study (which is limited to California), you’ll see that it reports on women staying in shelters for battered women. Women staying in battered women shelters aren’t just the typical battered women — they are women who feel so endangered that they decide to flee their own homes. It stands to reason that many of them would have been deliberately threatened with death or serious injury at some point, which may be what led them to flee in the first instance; they are probably more likely than the typical person, including the typical domestic violence victim, to have been threatened in an especially serious way. And it stands to reason that in those homes where a gun is present, and the woman’s partner is willing to deliberately threaten the woman enough to cause her to flee, at least one of the threats will have involved the gun.
So the study (even if it’s sound on its own terms, a matter on which I won’t opine here) doesn’t actually speak about battered women’s households — it just speaks about the households of those California battered women who felt so endangered that they go to a battered women’s shelter.
Mark Field says:
The study itself says:
“To obtain such information,
we interviewed residents of battered
women’s shelters—women who were
likely to be representative of those who have
experienced substantial amounts of violence
and who have had various objects used
against them by an intimate partner.”
I didn’t see any support for what seems to be merely an assumption of “representative”. This adds to the problem you noted.
It would be important to know (a) what percentage of battered women seek shelter; and (b) in what ways that percentage is representative (or not) of the total population. Mark Field(Quote)
Duffy Pratt says:
You can go back even further if you want. There’s no clear definition of a “battered woman’ here either. The numbers will likely expand or contract depending on how much violence a woman needs to have suffered, and what steps she must have taken in response to the violence, before she can be called “battered.”
Also, you may be right in your assumption that the numbers are inflated because they took the sample from woman in shelters. But its also possible that this underestimates the number as well. In my experience with the TRO project at Yale, there were a substantial number of woman who were afraid to get any sort of help at all because of the retaliation that might ensue. And there were more who took initial consultation steps, but were afraid to either leave the batterer or take any legal action against him, because they didn’t think the law would be able to protect them. So while the ones who fled to a shelter may represent one sort of extreme case, there are lots of other extreme cases where the women never get near a shelter. Duffy Pratt(Quote)
rarango says:
I fear that, in general, we are become less numerate as a society, and even less familiar with
underlying principles of research methodology. rarango(Quote)
zuch says:
Duffy Pratt:
Oh, you could ask her insurance company if she’d been denied for a “pre-existing condition”. Insurance companies are usually pretty good on the salient details, and drawing the appropriate lines.... ;-)
Cheers, zuch(Quote)
yankee says:
No doubt women in battered women’s shelters are unrepresentative, but it’s far from clear that they’re unrepresentative in the way you suggest. Most of the reasons women leave their abusers (or don’t) aren’t directly related to the degree of violence they’ve suffered—the extent of their support network outside of the marriage, the degree to which they think they’ll be believed, their financial dependence on the abuser, and the woman’s ability to take her children with her all play a huge role. And as Duffy Pratt says more violence may make a woman less willing to leave rather than more if she fears even more severe retaliation. yankee(Quote)
ptt says:
My goodness, misrepresentation — whether deliberate or accidental — on a legal academic discussion list, an unidentified one, at that!
Someone alert the authorities! ptt(Quote)
geokstr says:
I think “distorted” puts way too kindly a spin on what is done with “statistics” these days. They are often used by “researchers” to prove the conclusion they started with, to support a political agenda they have been pushing all along. Much of the useless mush that comes out of the various Victims Studies Departments in big name universities for starters. geokstr(Quote)
yankee says:
I don’t know much about the statistics that’s done there, but you’ve clearly studied it extensively so I will defer to your considerable expertise on this point. yankee(Quote)
LarryA says:
IOW in battered women’s households that contained a firearm, the intimate partner did not use the gun to shoot at the woman 95% of the time.
Most battered women who are murdered are beaten to death. LarryA(Quote)
tom says:
Of course, having just complained about drawing conclusions from studies that may use misleading statisistics, you Professor Volokh, proceed to draw a number of conclusions abouut why women in shelters are not representative of battered women as a whole.
What preciscely is your evidence that for your inferences in the following paragraph:
Is that not a bit of the pot calling the kettle black? tom(Quote)
Splunge says:
Pfui, they missed a clear chance to quote even more impressive statistics:
In over 90% of the households of female victims of assault with a firearm that contained a firearm, an intimate partner used the gun against the woman.
Scary!
Plus you can replace “90%” with any percentage you like, lower for the more suspicious audience, higher for the more credulous.
In other news, 100% of the victims of handgun violence were KILLED BY GUNS. Ban guns! Splunge(Quote)
Dudeman says:
Tom:
Note the difference between “conclusions of the study” and “conclusions drawn from the study.” Dudeman(Quote)
Allan Walstad says:
Two points.
1. It may well be that sociopaths are more likely armed than non-sociopaths. If the gun was in the house because a violent sociopath kept it, that says zippo about the likelihood of misuse of guns by normal law-abiding people. This goes back to the old confusion between A) homes where people were killed, assaulted, or threatened by a gun being more likely to have guns kept there, versus B) people who keep guns being more likely to be killed, threatened, or assaulted by guns. It’s possible (and likely the case) that A is true while B is false, because if your statistics start with cases of gun violence they are highly unrepresentative of the general population.
2. The interesting question would be what happened to women who armed THEMSELVES in reponse to violence and/or threats. Did the study distinguish between those cases versus others, where the PARTNER kept a gun? Allan Walstad(Quote)
Houston Lawyer says:
What I have to wonder about is how in the presence of so many guns and so much violence so few people actually get shot. Houston Lawyer(Quote)
Dotar Sojat says:
Ah, but the narrative. Toujours le narrative. Dotar Sojat(Quote)
TheBadness says:
Another anecdote serving as evidence for my proposition: a statistics 101 requirement in high school. Leave the advanced maths for college; throw basic stat at kids when the algebra is still fresh.
Stats class: it’s not rocket science, people. And it’s one of those great classes that the average person can use when they’re reading the morning paper! When was the last time organic chemistry made the front page? TheBadness(Quote)
ShelbyC says:
I wonder what the stats are for formerly battered widows with firearms in the home? ShelbyC(Quote)
theobromophile says:
Nothing against EV’s analysis, but it seems to miss the main problem with this statistic: it’s terribly circular. There is no control group (e.g. fire-arm owning households that don’t involve battered women). I suspect that something like 0% of non-abusive husbands in gun-owning households will have used them to threaten or shoot at their wives... but the Brady campaign doesn’t have that statistic for us.
Batterers, by definition, threaten to harm or actually harm their victims; that a batterer acted like a batterer isn’t really news. If his weapon of choice weren’t a gun, it would be a baseball bat, a knife, or a tire iron. So all this really seems to be saying is that battered women are battered women. (Maybe that’s more tautological than circular; I’ll let y’all decide.) theobromophile(Quote)
zuch says:
Houston Lawyer:
Are you aware of the entymology of the word “decimated”? That carries the context of horrible, unacceptable losses ... but it meant that, in a war situation with guns all over the place and people deliberately and unabashedly shooting at each other, the army had lost one in every ten soldiers. Does that put things in context?
Cheers, zuch(Quote)
tamerlane says:
I’ve done a fair amount of reading and research in this area and presented some professional papers on the topic based on analyses of the FBI’s SHR. I find this assertion hard to believe. (Actually I think it’s absurd.) Would you correct my ignorance by providing at least one citation from a reputable source tamerlane(Quote)
Strict says:
“What I have to wonder about is how in the presence of so many guns and so much violence so few people actually get shot.”
How many is “so few”?
I don’t know how many people got shot in 2008. I would guess around 100,000. Assuming that’s close to the actual number, would you really think that is “so few”? Strict(Quote)
zuch says:
theobromophile says:
It would seem that you miss the point that such threatening or shooting is arguably per se abuse (unless you’re of the curious opinion that sometimes it’s just sport). Kind of like saying that we find it highly significant that an astonishing 0 percent of watermelons are oranges.
Cheers, zuch(Quote)
Strict says:
“I suspect that something like 0% of non-abusive husbands in gun-owning households will have used them to threaten or shoot at their wives...”
If you threaten your wife with a gun, you aren’t “non-abusive,” so by definition 0% of non-abusive husbands have not threatened their wives. Strict(Quote)
Allan Walstad says:
Zuch & Strict: Try reading Theobromophile’s comment again. Try to understand it this time.
Theo–thanks for putting my point #1 a different way. They don’t have a control group because the statistical base they’re working from starts with cases of abuse. Allan Walstad(Quote)
Strict says:
Sorry, typo.
“so by definition 0% of non-abusive husbands have threatened their wives.” Strict(Quote)
tamerlane says:
You are an extreme example of what this post is directed against, i.e., the misuuse or confabulation of statistics. Actually 78% of women killed by husbands or other sexual partners are shot or stabbed — see Table 3. Since a large majority of the women who are not shot or stabbed to death are — by my recollection — strangled or garroted only a very minute fraction of battered women who are murdered are “beaten to death”. tamerlane(Quote)
Abdul Abulbul Amir says:
Here is another:
If true, this makes a gun kept at home approximately 10 times safer than a bathtub. This is true because a bathtub is about 500+ times more likely to kill a member of the household or visiting guest than an intruder. The same is essentially true for step ladders, marbles, buckets of water, stairs, steak knives, plastic bags, toasters, and just about everything else you find in a typical house.
The fallacy rests on the fact that killing intruders is not the reason to keep either a gun or a bathtub at home. Abdul Abulbul Amir(Quote)
Jmaie says:
Are you aware of the entymology of the word “decimated”? That carries the context of horrible, unacceptable losses ... but it meant that, in a war situation with guns all over the place and people deliberately and unabashedly shooting at each other, the army had lost one in every ten soldiers. Does that put things in context?
The word is from Roman times where, with no guns and no shooting whatsoever, 10% of a unit was bludgeoned to death as punishment for cowardice. It has nothing to do with unacceptable losses.
Not sure what the heck this has to do with the topic or Houston Lawyer’s comment. Jmaie(Quote)
JeffH says:
Here is what I came up with on short notice. The data is from the FBI’s National Incident-Based Reporting System.
The data below is the result of searching for reports that were classified as Domestic Violence where the most serious injury to the victim was a “Major Injury”. Major Injury is defined as Death, broken bones, loss of teeth, severe laceration, or unconsciousness.
The database contains 13,279 reports that meet those criteria. Of those, the type of weapon reported at the scene was:
Firearm: 1004
Knife: 2930
Blunt Object: 1598
Personal (Hands, Fists, Feet): 5551
Other: 1497
Unknown: 324
None: 375
The tool is available here if you want to look at it: http://ojjdp.ncjrs.org/ojstatbb/ezanibrsdv/asp/selection.asp
There are also pages about how the data is collected and how the search tool works.
The results above are from selecting “most serious injury to victim” in the “Row Variable”, “Type of Weapon” in the “column variable”, and “All Reporting States” as the State variable. JeffH(Quote)
uh_clem says:
Did you know that 84.3 percent of statistics are just made up on the spot? uh_clem(Quote)
jack burton says:
Best estimate there are about 360,000,000 guns in America. By your numbers, 359,900,000 didn’t shoot anyone last year. So yeah... in the presence of so many guns there really are so few people being shot. Somewhere along the line of less than one tenth of one percent. Would you really think that is a “large percentage”? jack burton(Quote)
ShelbyC says:
Um, I believe that’s the point, no? The inverse of showing that X% of abusive husbands have threatened their wives with guns? ShelbyC(Quote)
Pintler says:
The cited Brady page contains other interesting tidbits:
-“Firearms appear to be more common in homes where battering has occurred (36.7 percent) than in the general population (16.7 percent)”
[17% of houses containing a gun is much lower than e.g. Kleck’s numbers. Moreover, the source Brady cited by Brady gives contains the 36.7% but not the 16.7%]
-“Gun owners are 7.8 times more likely than non-gun owners to have threatened their partners with guns”
[this one just confuses me. Are the non gun owners threatening their partners with borrowed guns, or threatening to obtain a gun and then use it on the partner, or what?]
(on the education front, give copies of “How to Lie with Statistics” to your kids/nieces/nephews. No one should graduate from high school, or God forbid, vote without reading it) Pintler(Quote)
TruePath says:
I actually don’t find this misleading or problematic at all. The authors are clearly stating their assumption that these women are representative of those who have experienced substantial violence and informing the reader this is nothing more than a brute intuition on their part. While this might turn out to be false it’s a reasonable view (as is it’s compliment) to have in the absence of undermining information.
I mean this is exactly what scientists are supposed to do, clearly stating the assumptions their work rests upon. Even in hard sciences like astronomy or physics ultimately the result rests on brute judgements of plausibility by the author, e.g., an experiment to measure Hubble’s constant by comparing brightness and red-shift measurements of supernova simply has to make the brute assumption that we don’t live in a particularly dust free region of space. Making these kinds of intuitive judgments is the only way to avoid extreme skepticism so the best science can do is to clearly state them.
The problem really lies with those interpreting the study. TruePath(Quote)
Allan Walstad says:
Speaking of how statistics cam be misused, here’s a quote from the study, p. 1416, column 1, 2/3 of the way down:
The footnote is to a 1993 paper by Kellermann et al–I believe is the one that was totally debunked years ago. The actual result of THAT study was that homes in which people were killed were more likely to have guns. Anybody who can reason properly will understand that that does NOT mean having a gun in the home increases the likelihood of being killed. A study that starts from a database of crimes is highly skewed to the inclusion of violent sociopaths, and what happens to them has little to say about what happens to normal, law-abiding citizens who possess guns. Allan Walstad(Quote)
DianaWR says:
Part of the reason studies cover those seeking shelter services is, in part, because domestic violence is frequently hidden from sight until such time as either (a) a victim seeks shelter services or (b) the police get involved. Either way we are talking about the point at which the cycle of violence has escalated to the point that violence is imminent if not actually present. You know what typically precedes violence? Threats of violence. Seven years working in DV advocacy, I have to say I don’t find this sampling problematic given the rate at which threats become violence, the tie between the two, and the difficulty of identifying victims due to systematic shaming and isolation until violence is happening or about to happen. DianaWR(Quote)
zuch says:
Allan Walstad:
But what’s the probability that a home of normal, law-abiding citizens contains a violent sociopath? Yes, it’s quite comforting to know that if you’re in a home where there’s no one that is going to up and kill you, that therefore there is no one there that will up and kill you, but that rather begs the question....
Cheers, zuch(Quote)
Mark Field says:
It depends on the conclusions they draw. If they want to generalize conclusions to all battered women, then I think it’s incumbent on them to show the sample reflects that. If their conclusion is limited to “women in shelters”, then it’s no problem (obviously), but in that case I have to wonder why they brought it up. Mark Field(Quote)
zuch says:
Allan Walstad:
I did. I still have the same complaint as to his reasoning. I’m just a tad surprised that you are missing our point after having read it twice.
I have no idea what Theobromophile’s thinking WRT proper “controls” (or even what a “control group” actually is), but that ain’t it....
Cheers, zuch(Quote)
zuch says:
Jmaie says:
As I pointed out (and you confirm), it meant one out of ten dead. But carried forward, we still use the word decimated to indicate horrific losses even (or particularly) in the case of military conflicts. For instance, millions served in WWII and even in Vietnam, and only tens to hundreds of thousands lost their lives (there were some exceptions to this relatively low loss ratio, such as air crews in bombers and fighters over Europe, with staggeringly high loss rates). But such losses are still horrific. That was my point.
Cheers, zuch(Quote)
karrde says:
Gentleman, may I direct you to the National Institute of Health’s WISQARS tool?
Total deaths by firearm (suicide, homicide, police intervention, and accidental) in the year 2006 were 30896.
The number is much less than 100000.
The data remain within 5% of this value back to 2000.
During that time span, between 15000 to 17000 suicides by firearm are reported to the CDC per year, and 12000 to 14000 homicides by firearm. About 600 to 700 accidental deaths by firearm are recorded per year.
For comparison, during the year 2006 the deaths in the category Transportation-Related amounted to 49827. Most of these were categorized as Accidental, and were related to automobile transportation.
The estimates of firearms in private hands in America vary from 150000000 to 360000000; it is obvious that at least 99.98% of them were not used to end a life in any given year. karrde(Quote)
Allan Walstad says:
Zuch, it is sufficient to reply, in response to both your recent comments, by pointing out once again that a database built on crime cases automatically contains a highly skewed sample in relation to the population at large. So for example, if you start with a database of abused women, and then find that a certain fraction of the ones from homes with guns were threatened or attacked with guns, it can give the misleading impression that there’s a causal connection between gun ownership and violent crime, or that guns are more likely in general to victimize the innocent than to deter or be useful in self-defense.
Now it may be that the authors of the study were careful not to make such claims themselves. Yet how interesting that they do claim an increased general risk from firearms possession, citing another study, which in fact demonstrates no such thing.
The question-begging is being done by those who cite such studies to make claims about “guns as a risk factor.” Is it the gun that is the risk factor, or is it the sociopath? Start with sociopaths, draw conclusions about sociopaths. If sociopaths engage in violence, and if sociopaths have guns, the statistical correlation between guns and violence among sociopaths is simply a common-cause correlation where the common underlying variable is the sociopath. Yet anti-gun propagandists point to such studies to suggest that guns are a “risk factor” for everyone, that people in general increase their risk of harm by possessing guns.
By the way, I do find it rather comforting to know that I’m in a home where there is no one who is going to up and kill me, and where an outsider who entered with that purpose would encounter people prepared to defend themselves. Allan Walstad(Quote)
Strict says:
” Strict:I don’t know how many people got shot in 2008.I would guess around 100,000.Assuming that’s close to the actual number, would you really think that is “so few”?
Gentleman, may I direct you to the National Institute of Health’s WISQARS tool?
Total deaths by firearm (suicide, homicide, police intervention, and accidental) in the year 2006 were 30896.
The number is much less than 100000. ”
lol.
1. That’s DEATHS by firearm. We we talking about PEOPLE GETTING SHOT. People shot includes the 30896 killed AND those injured (maybe 70,000?).
2. That’s 2006.
3. Still the question goes unanswered: is the number of people shot really “so few”? Strict(Quote)
loki13 says:
Guns don’t kill people. Statistics do.
(as a side note, I appreciate the posts of the people who confined their comments to the shoddy statistics. Every post that happens to deal with guns is not just a wonderful opportunity to talk about the relative merits of guns for self defense vs. overthrowing tyranny.) loki13(Quote)
Jmaie says:
Zuch — I still don’t see what war time casualty rates have to do with battered women or private gun ownership. Jmaie(Quote)
Malvolio says:
I am curious about the 7,728 crime scenes at which hands and feet were not present.
I’m also curious about the over-all point of the study. Yes, abusive husbands do make violent threats but fortunately, usually don’t carry them out.
My guess: those women are not in shelters. They’re home enjoying the peace and quiet, while their quondam batters are in a shelter, the hospital, or the morgue. Malvolio(Quote)
mattski says:
Why don’t you scroll up and read the comment he was responding to? mattski(Quote)
tommyboy says:
I question any survey that is done with victims who may feel they have something to gain by giving the correct answer. I only trust survey results that agree with my pre-determined positions anyway :). tommyboy(Quote)
David Nieporent says:
Do you have a study to cite in support of that? David Nieporent(Quote)
DJR says:
“the typical battered women”
You know, the typical ones, those who’re okay with being beaten in their own homes and generally feel safe except for the occasional slapping around. The shelter battered women, well, they might have it bad but that’s not typical.
I’ll add this to my file of tin-eared Volokhisms that I will pull out only if anyone ever nominates EV for a judgeship. DJR(Quote)
theobromophile says:
Thanks, Allan. Yes, Zuch and Strict missed my point. It happens. I just wish that those who don’t quite get what someone is saying would not assume that the individual is saying the exact opposite. theobromophile(Quote)
theobromophile says:
Just saw this. Zuch, don’t take this the wrong way, but it’s pretty clear (like 100% clear) that I’m a woman. If you can’t even look at the avatar to figure out my sex, it’s hard to believe you when you say that you’ve read my comment with any real sense of understanding. Really hard. Like impossibly hard, in fact.
Just saying. theobromophile(Quote)
zuch says:
Theobromophile:
Oh, really. So what’s my gender, looking at my gravatar? Sorry if I misstated your gender, but it’s not that important. Using a strained sentence construction that allows use of the gender neutral “their” isn’t worth the bother. But just a note: Looking at your gravatar, it’s still not obvious that the person there is female. But that’s what you get with 32X32 or whatever it is, too.... BTW, your nick might have been a better clue ... even if I’m the one that loves chocolate in our household.
Cheers, zuch(Quote)
yankee says:
I suppose it is news of some sort that batterers who have guns in the house often use them to threaten their victims. Not that this is a surprise, but research that confirms what you’d expect to find can still be important.
Of course, the Brady Campaign’s use of this statistic to imply that something is wrong with gun ownership makes very little sense. yankee(Quote)
zuch says:
Theobromophile:
Let’s review:
To put it another way, hard to have a “control group” that (even by your guess) doesn’t even exist. I explained one reason why it doesn’t exist. We could also lament the fact that the statistics are also not present for those households where the abuser threatened the spouse with a gun but where there wasn’t actually any such gun present (a similar situation). Then we could see whether the presence of the gun makes a difference as to when gun violence was threatened or committed. Needless to say, though, this latter “control group” is also a sparse one. Do you similarly think that this statistic should have been collected? If so, why? If not, why not? Then think about what I said in my first reply.
Cheers, zuch(Quote)
zuch says:
Allan Walstad:
An amazing 99.9% of all drivers manage to make it though the day without hitting something. Therefore all accident statistics are based on a minuscule and irrelevant number of very rare cases, and all traffic stats are bogus. Granted that this small number is perhaps highly skewed in favour of drunk drivers. Which explains everything, right?
Has it occurred to you that the number of women threatened or shot with a gun in homes that don’t have a gun is rather small? Why is this? Could it be that gun possession is the sine qua non for threatening or shooting? Do you dispute this “casual connection” of gun ownership and use of a gun for such purposes?
It may be that sociopaths are just as likely to threaten or batter their spouse whether they have the gun or not (or it may be that gun possession makes it more or less likely). This is your assumption here. Feel free to prove it (and also prove your underlying assumption that such threatening or battery is done only by sociopaths). But what’s fairly easy to figure out is that such threats and battery are generally more serious when a gun is involved, because of the intrinsic lethal nature of firearms.
This is indeed a “suggestion”, and is amenable to statistical analysis. Go for it.
I’m glad you “know” this. It’s warm comfort to all those in similar situations too. Problem solved, eh?
Cheers, zuch(Quote)
Tamerlane says:
JeffH:
The original assertion with which I disagreed was “Most battered women who are murdered are beaten to death.” THE SHR data presented in my citation clearly demonstrate that this is false. The SHR — precisely because of the seriousness of homicide — is a fairly reliable and complete set of data on homicides in the US. There’s no reason to doubt the essential correctness of data from this sourcer.
The NIBRS data you present just show that most women who are badly hurt by their sexual partners are beaten. This doesn’t contradict the SHR data. It’s also worth noting that there are still serious issues with NBRS reporting. There’s an extensive literature on this. It essentially boils down to the fact that NIBRS reporting requirements are so onerous that most police agencies opt out of the system. Tamerlane(Quote)
Allan Walstad says:
Zuch:
Don’t you think it’s a safe assumption that very few of the people who get in traffic accidents actually intended to do so, actually intended to cause damage and injury? The opposite is true with assaults. A salient difference, no? Nor am I saying that statistics, such as those under discussion regarding battered women and how many were threatened with guns etc, are bogus per se. I’m saying that they can very easily be misinterpreted as demonstrating a high general risk from legal gun ownership in society at large. In fact, the authors of the study under discussion themselves so misinterpreted the results of an earlier published study (as I discussed above).
No one is disputing that guns are weapons that can hurt or kill. They can be used for assault by sociopaths, and they can be used for self-defense by others. What I’m pointing out is that a database that starts with assaults is a database highly unrepresentative of the general population in salient respects. The conclusions you might draw therefrom, regarding the general population, are limited in essential ways.
Zuch, imputing assumptions to others and attempting to shift the burden of demonstration are rhetorical motifs that wear thin quickly. What I’ve pointed out is how statistics that start with violent crime can, and are, misinterpreted with regard to the risks or benefits of firearms possession generally. When, decade after decade, statistical studies are published that lend themselves to such misinterpretation, and are regularly picked up by the media and so used, it does frankly get rather annoying. As far as the more general discussion of guns is concerned, that keeps coming around but this particular thread is about done. Allan Walstad(Quote)
zuch says:
Allan Walstad:
Not in this case. My point was that the relative infrequency of an act (such as threatening or battery in households with guns) doesn’t make statistics concerning such relatively rare acts bogus (nor the acts themselves less serious). Nor does does the infrequency undermine any causal relationships (if any) that might be in play.
Who said it did? But if it were to — say — double a very small risk, wouldn’t that be significant?
A general population that doesn’t have sociopaths (or jealous husbands, or first-class jerks, or .... well, you get the idea). Of course, it is representative of a general population that does have a small share of such....
FWIW, we are not a very adept society at assessing relative risks and costs. That doesn’t mean, however, that we shouldn’t actually try....
Cheers, zuch(Quote)
Jack Burton says:
Compared to the total number of guns... yes. Which was the point the original poster made and which you are ignoring over several posts. Jack Burton(Quote)
Pintler says:
That depends on what you mean by significant. If one uses ‘significant’ to mean ‘this is an area on which society should focus its attention’, then no.
Let’s say you face a risk A of 10 in 100, and you can spend effort X to reduce it a tiny bit — to 9 in 100, or you can spend effort X to halve risk B, from 2 in 100000 to 1 in 100000. Spending the effort on risk B is a very non optimal choice.
Should we spend our limited resources on better guardrails, increasing the immunization rate, more community policing, banning private sales of guns, or arresting streakers wearing pumpkins as hats? You can’t do everything, so you have to pick what matters most. Merely knowing some proposed policy results in benefits isn’t enough — you need to compare it to the possible benefits of alternative policies. Pintler(Quote)
zuch says:
Jack Burton:
I’d say that it’s cold comfort to you to know that even a thousand well-mannered and law-abiding gun owners out there are no threat to you ... when your abusive hubby can easily go out and get a gun. If they were allowed to go out and get an absentee ballot or a constitutionally protected blue movie DVD, there wouldn’t be quite the concern. But you’ll have to admit there’s something inherent in guns that makes them just a tad different.
Cheers, zuch(Quote)
zuch says:
Pintler:
That depends on what you mean by significant. If one uses ‘significant’ to mean ‘this is an area on which society should focus its attention’, then no.
I think that the abused wife might beg to differ. While the vast majority of wives aren’t battered, the increased risk for the few that are is rather significant in their eyes. It may be that you are of the opinion that society ought to say, “there’s not too many; screw them, I have to have mine”. But opinions vary on such things.
Cheers, zuch(Quote)
Allan Walstad says:
Nor did I say, Zuch, that the relative infrequency of an act makes statistics concerning such relatively rare acts bogus. My point is about the limited conclusions you can draw relative to the overall population from a highly skewed sample. It’s quite possible that people who get in car accidents are a reasonably fair cross-section of drivers. It’s not reasonable to take wife-batterers as a cross-section of men, nor households with wife-battering as a cross-section of households.
That’s how I took your reference to car accidents, with your apparently sarcastic statement that “all traffic stats are bogus” as analogous to my criticism of the misuse of crime statistics. Sorry if I misinterpreted.
Surely not statistically representative. But yes, if 75% of sociopaths have guns and 1% of the population is sociopaths (just making up numbers to illustrate), then 0.75% of the population is sociopaths with guns. You can draw statistical conclusions or at least estimates or bounds of that sort. Yes, the violent sociopaths are part of the overall population. If that’s your point, then the point is taken and was never in dispute. But if they are a small part, the statistics relating to them do not necessarily characterize the overall statistics.
More importantly, anti-gun propagandists often take the statistics from these skewed samples and use them to try to convince the rest of us that they apply to us. So, we are told that guns are a risk factor, that their possession increases the risk of injury and death, that WE put ourselves in danger by owning guns. (And I take it they are appealing to the thoughtful conscientious law-abiding among us, as well as policy-makers, when they do that, rather than somehow thinking that violent sociopaths will be persuaded lay down their arms.) But the statistics they appeal to show no such thing. No statistical connection between gun-possession and violence by violent sociopaths demonstrates that the rest of us put ourselves and loved ones at risk by possessing guns. Allan Walstad(Quote)
zuch says:
Allan Walstad:
Both these statements are ... at the very least, unsupported. Some might say suspect.
Cheers. zuch(Quote)
zuch says:
Allan Walstad:
[zuch]: A general population that doesn’t have sociopaths (or jealous husbands, or first-class jerks, or .... well, you get the idea). Of course, it is representative of a general population that does have a small share of such....
[Allan]: Surely not statistically representative.
You have a very strange idea of what “statistically representative” is, then. It does not correspond to what statisticians (as opposed to polemicists, for instance) would say.
I’d note that the database of those that were not assaulted that were also not assaulted with a gun is a slim one indeed. This is a concept that seems to escape you here. There’s no point in asking what percentage of those that were not assaulted (your “general polulation”) were not assaulted with a gun. Why you seem to lament this missing piece of data is beyond me. Capece?
Cheers, zuch(Quote)
theobromophile says:
Zuch: so my gender isn’t important but, even in a cocktail dress and with hair down to my waist, I look like a man? Are you trying to be a jerk, or was some humour lost in translation?
As for the actual substance of this discussion: my point was simply that the statistic itself is meaningless — partly because there is no control group and more so because it’s more about scaring people than about investigating the relationship between guns and domestic violence. As I said before, 0% of non-abusive husbands have threatened to shoot their wives. 100% of gun-owning, non-abusive husbands and boyfriends have firearms in their homes and do not threaten women with them. theobromophile(Quote)
Allan Walstad says:
Zuch: What I mean by “statistically representative” is what anybody having any acquaintance with statistics will tell you it means, namely, that if a statement is true with respect to x% of the sample then it is true (to some reasonable accuracy) with respect to x% of the general population. For example, if roughly half the population is male and half female, then a sample consisting of 90 males and 10 females is not statistically representative, at least not with regard to gender, because it is not true of the general population (as it is with the sample) that 90% are male.
Apply this to households with battered wives. Of those, look at the households containing guns. In some fraction of those households, the gun was used to threaten or assault the wife. Can we conclude therefrom that in the SAME fraction (or any remotely similar fraction) of gun-containing households in the general population, the gun has been or will be used to threaten or assault the wife? No. The statistical sample was selected precisely on the basis of there being violent abuse. It is not a statistically representative sample in that respect. For all I know, it might be a statistically representative sample with respect to the number of different brands of cars they drive or coffee they drink, but that’s not what we are talking about. More generally, a statistical sample that is selected in such a way as to include a disproportionate number of violent sociopaths is unlikely to be representative of the general population with respect to gun use or abuse or victimization. Or at least, there is no reason to believe that it would be representative in that respect.
That’s the best I can do. The fact that you don’t understand what is meant by “statistically representative” indicates that you are not prepared at this point to engage in a substantive discussion of the interpretation of statistical studies. Allan Walstad(Quote)
zuch says:
Theobromophile:
As I said before, threatening to shoot your wife is per se abusive, so this “statistic” is irrelevant; a statistical tautology.
As to your gender, I’m serious: The gravatar hardly does you justice, had I even bothered to look at it. This may not be your fault; the resolution is horrible. And also: my hair was until very recently almost as long.
Cheers, zuch(Quote)
zuch says:
Allan Walstad:
Actually, it’s the converse of the bolded statement. The population statistic is not uncertain or variable. It makes no sense to talk of what is true of the general population “to some reasonable accuracy”. It is the estimate of the population statistic (from the sample statistic) that has some uncertainty, but that doesn’t change the actual population statistic.
Your problem here is that threatening or battery with a gun is per se assault or battery. So any such posited instance of such a threat or assault in your “general population” automatically places that instance in the more limited group of “households with battered wives” (or such similar situation). By definition, you won’t have any such instances in households with non-battered wives, whether or not they have guns. How can such a group be an adequate (much less a necessary) “control”?
But I’d note that no one as claimed any such “SAME fraction”. That’s just you making facts up.
Cheers, zuch(Quote)
uberVU - social comments says:
Tweets that mention The Volokh Conspiracy » Blog Archive » How Statistics Get Distorted — An Example -- Topsy.com says:
jack burton says:
yes, some people do horrible things with their freedom that hurt other people. And if one is the person getting hurt I understand that it is perhaps meaningless to them at that moment that millions upon millions of other people are not being hurt. But we make our laws on logic and reason... not on the passion of the moment and high levels of emotionalism.
When you make laws that negatively affect the freedoms of those 99.99 percent of people who are doing no harm... it becomes a great concern. jack burton(Quote)
zuch says:
jack burton:
So, as I essentially said above to someone else, it’s like this: “Oh, it’s just a few whiners who are bitching about being put in mortal danger. Just a tiny number, though, so f**k ‘em, I got mine....”
Cheers, zuch(Quote)
jack burton says:
And that is the best you can do, eh... jack burton(Quote)
theobromophile says:
zuch: my point, again, is that the Brady statistic is just as meaningless as the one I gave. (Yes, I’m well aware that threatening a person with a gun is abusive. That was my point.) Allan (and other commenters) pointed out that abusers always threaten or use some sort of force (beatings, knifings, etc); the Brady statistic thus becomes tautological. theobromophile(Quote)