Hacker News Reader

Comment by Empact | Hacker News Reader

Late last year in response to IMO misleading charts showing a clear correlation between "gun deaths" vs gun freedoms, I charted murder vs gun rights from a city and state perspective, which illustrate some of the points made in the article. Helpful to visualize, e.g. the point made about Louisiana as an outlier:

https://hackpad.com/Gun-Rights-Statistics-UBI1bkaNgG6#:h=No-...

Raw data and interactive charts here: https://docs.google.com/spreadsheets/d/1N5JsB-_kTxFSW-14f7Bv...

View on HN

Very cool, thanks for sharing.

I think it is interesting that, with the exception of Hawaii, all of the lowest murder-rate states are northern, mostly rural, and overwhelmingly white. I don't know what conclusions to draw from that, aside from the general trend that these aren't states where there is a lot of ethnic gang violence.

Murder is highly correlated with urban and minority populations - the murder rate for blacks is about 8x that of whites.[1] The root of this is clearly socioeconomic, and I believe rooted in the drug war[2], which drives black market violence both by creating a lucrative ungoverned market and by removing constructive inter-generational alternatives through mass incarceration - i.e. if your father is in prison for a non-violent drug crime, you as a child lose both a role model and a bread-winner, which makes you more likely to fall into gangs for social and economic reasons.

If that's accurate, it's a vicious, massively destructive cycle.

[1] https://en.wikipedia.org/wiki/Race_and_crime_in_the_United_S...

[2] e.g. outlier city "Baltimore’s City Police Commissioner Anthony Batts says that 80 percent are drug-related." http://www.thedailybeast.com/articles/2014/01/01/the-year-in...

"Eighty percent of the homicides were gang-related, [Chicago Police Commissioner Garry McCarthy] said" http://www.huffingtonpost.com/2012/12/29/chicago-homicide-ra...

New Orleans Police Department: "29% of all murders in #NOLA this year involved a gang member. That's down from 42% this same time last year." https://twitter.com/nopdnews/status/618864980364898305

And in some California cities "gang homicides only accounted for 29 percent of the total for the period under consideration (2003-2008)" I question the use of "only" here. :P http://www.huffingtonpost.com/evan-defilippis/do-we-have-a-g...

On the other hand, The National Gang Center says (based on survey data): "These estimates suggest that gang-related homicides typically accounted for around 13 percent of all homicides annually." https://www.nationalgangcenter.gov/survey-analysis/measuring...

It's hardly clear that the root is socioeconomic. I don't have data on crime, but in education race remains predictive even after you include socioeconomic status in the estimator.

https://randomcriticalanalysis.wordpress.com/2015/05/16/on-c...

Do you know of a similar analysis of crime?

Agreed. I've looked into the data in the past and this is not at all clear. You don't see anywhere near the same incidence for murder or violent crime among impoverished whites or Hispanics in America. These sorts of things are much more correlated with IQ than to poverty or nearly anything else.

I haven't looked into this deeply, more of a productive assumption[1], but here are some related studies:

"the majority of the black-white gap (over 60%) [in violence] and the entire Latino-white gap are explained by a small set of factors, especially marital status of parents, immigrant generation, and neighborhood characteristics associated with racial segregation." https://www.rand.org/content/dam/rand/www/external/labor/sem...

"Despite a large difference in mean levels of family disruption between black and white communities, the percentage of white families headed by a female also had a significant effect on white juvenile and white adult violence."

"The combination of urban poverty and family disruption concentrated by race is particularly severe. Whereas the majority of poor blacks live in communities characterized by high rates of family disruption, most poor whites, even those from "broken homes," live in areas of relative family stability" https://dash.harvard.edu/bitstream/handle/1/3226952/Sampson_...

"Multivariate regression results for ninety-one cities showed that while total inequality and intraracial inequality had no significant association with offending rates, interracial inequality was a strong predictor of the overall violent crime rate and the Black-on-Black crime rate." http://egov.ufsc.br/portal/sites/default/files/anexos/33027-...

[1] I have a hard time imagining a productive line of thought that leads from saying that high black murder rates result from something inherent to blackness

When reading studies like this, it's very helpful to look at the tables of regression coefficients rather than just reading text. Unfortunately, it can often be a career limiting move for study text to accurately reflect the contents of the data tables/regression coefficients.

The first source you cited (the rand study) shows that being African American is a strong predictor of engaging in violence, even after accounting for other factors. See table 2.

The second study doesn't address the question.

The third study (see table 2) shows that, among other factors, percentage of a city that is black is a strong predictor of violent crime even after accounting for other factors.

One productive line of thought which leads from "something inherent to blackness causes crime" is "since this problem is intractible with current levels of biotechnology/social engineering/etc, we should stop wasting resources trying to solve it and focus on other things."

I don't disagree with your analysis of the studies, but the fact that a difference remains after accounting for known socioeconomic factors does not indicate that other factors do not exist.

Personally, I wouldn't call this[1] a productive thought, rather a fatalistic one.

[1] "since this problem is intractable with current levels of biotechnology/social engineering/etc, we should stop wasting resources trying to solve it and focus on other things"

This fairy tale about "non-violent drug crime" is interesting because it tries to get us to forget how many violent criminals are locked up THANKFULLY due to the only charge that stuck to them, e.g. their 9th felony possession charge while armed with a loaded gun. People would have us all believe that these peaceful minorities are locked up because the cops stopped them on their way to church, found a little "medicine" on them and put them away for 20 years and left his kids orphaned. It's just nonsense. So many people get leniency in the cases of repeated drug offenses. Honestly, what is a society supposed to do with people who repeatedly break laws?

In major cities a frightening number of homicides go unsolved. Though I can't say I'm a fan of the drug war, there is value in locking up criminals that just haven't been caught doing their most violent acts. Any time somebody brings up the non-violent drug imprisonment, and the socio-econmic downward spiral, yes it's true there are a small number of people who have been railroaded who could be raising their sons and daughters, but we also need the clear image of the killer who lucky for us got pinched because he made a different mistake. All the cops in the room know he killed his cousin but everyone in the project last year was too scared to rat him out. The prosecutor is going to push the longest sentence they can get for the laws broken just so that guy can't victimize somebody else's kid.

> there is value in locking up criminals that just haven't been caught doing their most violent acts.

you're making judgments before proof or evidence aka prejudice.

NO. I'm saying lock up somebody for one crime you have proof on, might not have been that criminal at their worst but you got them on the proof of another. They are paying the price and they can't do their worst crimes to innocent people when they are locked up for the moment. That's value to peaceful society.

I'm typically skeptical of conclusions like that, to be honest. If we look at the gun deaths graph, many of the highest rates are in much whiter, much more rural states, and the south is very well-represented.

I'll probably draw criticism from what I'm about to say, but perhaps part of this is due to cultural differences in what constitutes murder between different states. An obvious example would be the case of George Zimmerman and Trayvon Martin. I would classify that as a murder, but I don't think it would be counted as such in this data.

I think about these things a lot, as a minority who grew up in the most urban part of the country. It's very common for statistics to be misleading in what actually goes on in these areas for a number of reasons. I'm not saying that there is no such thing as a dangerous area with high murder rates, just that it's not always as simple as looking at some numbers.

I should also point out that while many high-crime areas have a high minority population, I would consider most of them rural or suburban. Milwaukee and Detroit for example, are not exactly "urban" cities.

I would classify that as a murder, but I don't think it would be counted as such in this data

Generally, the number used is homicides with a gun -- that encompasses self-defense, manslaughter, police shootings, etc.

I don't think justifiable homicide such as self-defense should count if you're wanting to answer a question of "should we keep guns or get rid of them"

Obviously having more justifiable homicides with guns is better than gun murders, because it's good people protecting themselves from violent people. I don't think a criminal who died from trying to kill me deserves to be added to a statistic to further an idea that "less guns are better"

Speaking of Trayvon Martin, we'll never really know what went down because we can't get in those two's heads, but I can tell you if I was armed and found myself getting my face beaten into the concrete, not knowing what the person would do next, I think I'd certainly feel justified in using a projectile weapon. If somebody is beating you senseless on a dark night, I don't know about you, but I wouldn't assume they'll just leave you after you black out. They may curb stomp you to death, they may take your gun and kill you, they may take your gun and kill somebody else.

I don't think it's a cultural difference between states, I think it's a legal line drawn between homicide vs justifiable homicide.

I'd like to see charting of both, because justifiable homicide is an indicator of danger in an area too. Unsuccessful attempts at violent crime resulting in justifiable homicide by firearm.

That's one obvious problem is you know that any partisan journalist can take justifiable homicides, lump them into the same bucket under gun deaths by murder and use it to reinforce a message of guns = violence. It's disingenuous.

Same thing with other crime. An area could say rapes decreased, even if attempted rapes went up. So it seems successful self-defense literally and figuratively means you won't become a statistic. Which is bad when you want an accurate picture of the safeness of an area.

No buyers agent can say, "Oh yeah this area is really safe, hardly any crime happens to people here, so long as you're open carrying a glock 19 and are an expert in Krav Maga."

The conclusion I think one can draw from your charts is that increased gun control does not correlate with reduced murder-rates.

So, anybody who claims to support data-driven decision making--you know, the A/B testing so in vogue right now--should find themselves having to say "Yeah, well, okay, this is really a non-issue".

Instead, we see continued political sniping and popular manipulation and twisting of the very idea of certain inalienable rights, all done for political and not numerically-proven policy reasons.

Sigh. :(

Ah, the Villain Effect. What the other guy says must be evil, so the opposite of that must be good, right?

Between that and the way we select politicians for it, it's a wonder there hasn't been a second civil war yet.

As a fan of logic I appreciate your post and understand the Sigh all too well.

None of this data includes accidents, or other effects like a more armed populace means a more armed police force and the every escalation of violence that entails.

The real comparison is not among states with pours borders, but from the US to other western nations with outright bans. Chicago for example is in a strict gun law state but is right next to Indian with lax laws. There is no border enforcement at the state level so it seams dubious that the Illinois laws have any effect. Our rate is about 3.8 times the rate of the UK and its hard to see at a macro level what other than the ubiquity of firearms is the difference.

Statisticians have frequently noted the difficulty in making such comparisons. Many European countries are much more homogeneous than the U.S. Also, you mentioned the porous borders between states reducing the effectiveness of said gun legislation. We have a fairly porous border with Mexico. Mexico currently has more drugs and people than firearms, but there's no reason we couldn't see a marked increase of arms smuggling into the U.S. in the event of stricter gun laws. In many ways it is more straightforward to manufacture firearms than meth, for example.

Accidents are not the same as intentional homicides, and should not be treated as such.

You do have a point re: arms races between the police and the rest of the country, but I'd argue that letting the police up the ante is a mistake. It's better for both sides to be armed equally, as this discourages officers from abusing their power (and vice versa).

As for lumping states together, there's a reason you don't do that, and it's codified in something called the Scientific Method. If you want to measure the effects of a certain well on iklness, you must count the users of that well separately from users of other wells, even if the wells are right next to each other. If you lump two wells together, then you don't know which well actually caused the illness.

Well only one kid was murdered in my high school but 3 died from accidents. We worry and track automobile accidents so why not gun accidents its all part of the picture.

My point is that the measure should be how far a city is from a state with lax laws not just which state it is in.

I would argue that banning firearms is a highly inefficient way to deal with firearms accidents. Accidents are a direct result of ignorance. You tell newly-licemsed firearms owners that 3 times as many kids die in firearms accidents as thdy do in shootings, and they will (hopefully) keep an eye on their kids.

As for how far a city is from a state with lax laws: Wouldn't the laws in the city itself have an effect as well? Some citizens follow the law, after all. One of the major arguments for free access to weaponry in general is that those who rob and kill and rape are usually willing to risk extended jail time in exchange for being able to use a gun to keep others from fighting back. The thing about threatening someone with a weapon is this: if they have a weapon, or a random bystander has a weapon and decides to interfere, then instead of forcing someone to cooperate, you've made the situation more dangerous for yourself (and everyone else).

Or because (1) the mechanisms for enforcement are weak at the state level, (2) because you're not controlling for other factors that increase crime and (3) the fact that massive gun violence has a direct causal link to more gun laws, you could reject the idea of state stats based on their "gun laws" being very meaningful and look at national stats, for example:

http://www.realclearscience.com/blog/gunsnosa.jpg

How about a useful state stat:

http://www.motherjones.com/files/gun-ownership-vs-gun-deaths...

The correlation disappears (when comparing the 50 states, or all countries, or all OECD countries) when you look at all homicides, not just homicides by shooting. In other words, there is no correlation between guns per capita and overall homicide rate.

Reply on news.ycombinator.com