Homicide statistics, wtf?

by Doctor Science

I got statistics for the previous post on Intimate Partner Homicide (IPH) from the Bureau of Justice Statistics Homicide Trends Report. I see that report comes from Uniform Crime Reporting Program Data Series, which I don't know how to massage.

But I wonder how reliable that data really is. In our discussion, Jason Kuznicki linked to 2003 paper about the possible role of changing to unilateral divorce laws. In discussing the data they used, the authors state:

One outlier to this is Illinois where the Chicago Police Department failed to report any murders in 1984, 1985, November 1986–May 1987, July 1987–
December 1987 and July 1990–December 1990. As it is implausible that there were no murders during these periods, we omit Illinois from our homicide samples.

I have no idea if these holes are there in the data used from the Homicide Trends Report, or if they're significant for other cities, states, and years.

I'm pretty sure, too, that somewhere in my browser history there's a paper about IPH that was using a figure of 300-and-something for a recent year, which is also wtf.

Basically, I get the feeling that the US homicide stats may have some serious holes, but I don't know enough about the field to know where/why/how. And everyone agrees that homicide is the crime for which stats are most accurate.

I was googling for images of "counting house" when I found this: wool samples and ledgers from Fox Brothers archives, or, to be more accurate, the attic.

What looking for statistics feels like. And I'm allergic to wool.

Meanwhile, I went looking for international comparisons. Here we run into a problem: the US has an extremely large population (3rd in the world) with a high homicide rate, so the total number of homicides (more than 15,000/year) is large enough for all kinds of statistically significant analysis. The countries that are closest to the US culturally are both much smaller and have a much lower homicide rate, so the number of homicides they experience makes for much less robust statistics. Canada has fewer than 600 homicides/year, the U.K. has fewer than 800. Countries that have a total number of homicides comparable to the US -- Brazil or Mexico, for instance -- aren't culturally very similar.

So I've flailed about for a bit, trying to find data from either the UK or Canada that can be compared to the US trends in IPH, but failed.

If anyone has any better links or more authoritative insights, cough 'em up. All I can say is that, almost everywhere in the world, the overall murder rate has been falling for a decade at least.

Comments

[Turbulence]

Hmmm...I'm at work right now, but I poked a bit at the 2009 homicide data dictionary and it seems reasonable. The data itself might be full of holes though. I'll look into massaging it tonight or maybe over the weekend; I'm especially curious about joining it with census data to get a better feel for how rates (rather than counts) change for different definitions of subpopulations.

[Doctor Science]

Turb:

Thanks! I poked at it, trying to get it to cough up a victim race/sex by offender race/sex chart for intimate homicides, but it is clear that I don't know what I'm doing.

The question I'd like to answer is: what percent of IPHs are intraracial, opposite-sex? That is, can we assume that almost all IPHs of white women are by white men, of black men are by black women, etc.?

[russell]

Countries that have a total number of homicides comparable to the US -- Brazil or Mexico, for instance -- aren't culturally very similar.

IMO this raises the question of what we mean when we say 'culture'.

How similar are the UK and Canada to the US in terms of culture?

How different are Mexico and Brazil, from us?

IMO it depends on what you are using as your cultural markers. There are a number of metrics by which we are not particularly like, frex, the other OECD countries who are typically cited as our peers.

[wj]

russell, we appear to know one thing: their cultures are dissimilar enough from ours (whatever the other similarities) that they don't have the murder rate that the US does. What part of their culture that is, however, is another question.

Could be the culture's attitude towards gun ownership. Could be the diversity of the population (generally, or in specific). Could be the attitude towards violence in general. Could be the attitude towards intimate partner interactions. But something is clearly different to produce such different results.

[Brett Bellmore]

One of the problems with determining that another country is culturally comparable to the US, is that the US is a large and culturally heterogeneous country. Much more the latter than a lot of first world nations. With a murder rate which varies by several orders of magnitude from place to place.

So, we might say in some very general sense that the US is "culturally comparable" to Canada. But is the part of the US where the murders are taking place culturally comparable to Canada?

[Sebastian]

Similar to Brett's point, we find artifacts like the fact that Minnesota has lots of similarities to Sweden (population health, inequality levels, education), while Michigan doesn't despite the fact that they are both big great lakes/trading history states.

[Brett Bellmore]

uihgb
vbml

[Brett Bellmore]

pbs
pbs
pbs

[Brett Bellmore]

Heh, my 3 year old got at the computer. ;)

The relevant cultural differences aren't on the scale of nations, or states. Hardly even on the level of cities. In many cities a few minute's walking can take you from fair safety to mortal danger.

We are talking cultural differences, but pretending they're on the level of countries is a joke, when it's not a ploy to use the crimes of one sub-culture to justify leaning on a different one.

[Brett Bellmore]

Incidentally, I once saw a graph, (Can't find it again!) which showed the percentage of over-all violent crime due to neighborhoods with different crime rates. It was frankly amazing how much of the crime was due to a tiny minority of outlier neighborhoods with crime rates hundreds or thousands of times higher than most of the population were subject to.

[Countme-In]

"pbs
pbs
pbs"

A budding Public Television Fan in the works?

Counter-revolution strikes.

[Countme-In]

Long term historical murder rates in the U.S. and Europe track each other sharply downward through the centuries:

http://marginalrevolution.com/marginalrevolution/2011/06/long-term-trend-in-homicide-rates.html

Couldn't find Canada for the same period of time, but why would it be different?

If we overlaid a graph of cultural heterogeneity over the murder rate per thousand, wouldn't it show high indications of homogeneity back when we all looked alike but apparently were knocking each other off willy-nilly but increasing heterogeneity as we've become more culturally diverse even as murder rates decline to the present time?

[Hartmut]

I think the heterogeneity is a very complex issue that can go both ways. There are numerous examples of very heterogeneous societies that lived in internal peace but rather suddenly the different groups were at another's throats and not always because a gifted demagogue came to town. And other groups that were thought of as never being able to live together with a long history of mutual violence somehow managed to do it (although this transition usually takes far longer). The same groups in one state (as in nation, not US subdivision) may have no conflict while in another they are mortal enemies. Differences can be huge multipliers for other factors not directly related.

It gets even more complicated when three groups interact. Historically one can find examples of Christians, Jews, and Muslims with two of the groups ganging up on the third in all possible variations.

I think in ultra-heterogeneous communities the deciding factor would be a general 'consensus' on the topic of violence, i.e. violence would be less based on and caused by the individual differences but by general mentality. Once there is a certain degree of segregation, i.e. people of the same kind flocking together, the groups would develop internal dynamics that can be violent or non-violent but limit themselves mainly to the group but also an attitude of us/them pointing outward that can become the seed of violence. Total segregation on the other hand (as long as the groups are comparable in size, so it does not resemble a ghetto or apartheid situation) would likely diminish negative inter-group relations or limit them to the border*. The main factors then would be intra-group.

*of course there are counterexamples galore

[russell]

But is the part of the US where the murders are taking place culturally comparable to Canada?

What part of Canada are you talking about?

Are you thinking that Canada, or the UK for that matter, are homogenous societies? Or, that there is not a geographical disparity of murder (or other crime) rates?

IMO there are large cultural differences between the US and other countries that we generally consider our peers - Europe, or the OECD countries in general.

But geographic and/or demographic disparities in crime are probably not one of the differences.

Americans kill - specifically, shoot - each other in statistically large numbers. IMO part of the reason for that is simply that there are more guns. Gun ownership is more common here, and personal ownership and use of guns for self defense, as opposed to (frex) hunting, is widely accepted as normal and legitimate.

The pros and cons of that to the side, if there are more guns around, it's relatively more likely that folks will get shot. All other things being equal.

Which, of course, they are not.

Other differences that are IMO worth noting are wealth disparity, the quality and availability of basic public services, and social and economic mobility.

And, it's always possible that we are, as a culture and a society, simply more violent than other people are. Or, at least, more tolerant of violence, and more willing to accept it as a legitimate way to resolve conflict.

There might be a chicken and egg question there - does the prevalence of violence due to other causes give rise to its social acceptance, or vice versa? - but to the casual observer named russell, it seems like violence and the use of force generally is not tolerated to the degree that it is here in countries that are otherwise considered our peers.

[Rob in CT]

Incidentally, I once saw a graph, (Can't find it again!) which showed the percentage of over-all violent crime due to neighborhoods with different crime rates. It was frankly amazing how much of the crime was due to a tiny minority of outlier neighborhoods with crime rates hundreds or thousands of times higher than most of the population were subject to.

I think the basic issue you've touched upon is that, in the aggregate we're a "1st world" country, but if you get granular enough what you will see is a mixture of 1st world with 3rd.

I can't prove any of this, and indeed this could just be me talking right out of my ass, but here's my general theory of crime:

High crime occurs when you have entire areas in which the people just don't buy in to society. They don't believe (rightly, wrongly, for whatever reason) that acting the right way (as the rest of us see that) as likely to pay off for them. Basically, it's a rejection of the implicit social contract (or rather the belief that the social contract does not apply to them, or would never be fairly applied to them).

I think a lot of upstanding, moral people are such basically because they believe/know that it will work for them. I beleive this is called by some "bourgeois morality" or middle-class morality. I think these people often fool themselves into believing they'd be just as morally good if their circumnstances were different. I certainly like to believe I would still be a mostly law abiding guy, but honestly? My whole life I've been certain in the belief that if I just played by the rules, my life would work out well (and it has).

Anyway, I see this is a two-way street and, therefore, a really difficult chicken & the egg problem. How do you convince people to buy in when they don't trust you? How do you convince the rest of your society to invest in (and sustain!) a non-"law and order" approach given the crime rates (which justifiably horrifies, confuses and angers them)?

[hairshirthedonist]

Basically, it's a rejection of the implicit social contract (or rather the belief that the social contract does not apply to them, or would never be fairly applied to them).

Considering the ages of people involved in a good number of the homocides that occur in my metropolitan area, I'd say there may be little awareness of the very notion of the social contract among a good number of the killers. That is, unless you take the simple choice between 1) *staying in whatever crappy school is available to you and/or getting whatever crappy job is available to you and* 2) *joining a drug gang, which seems to be the only way anyone around you gets money or power* as a choice between accepting or rejecting the social contract (which is probably not an invalid formulation, but one that I felt the need to make less abstract).

But, AFAICT, high-crime neighborhoods are places where lots of people have guns and lots of people have beefs with each other, mostly stemming from turf wars or other disputes related to the drug trade. They mostly kill each other, but not-at-all rarely innocent bystanders caught in the crossfire as well.

There are also "problem bars" doing their fair share.

As an aside, beyond the social contract-rejection question, I don't think there's very much of people being randomly killed by total strangers simply because they wandered into the wrong neighborhood. Certainly, some places are more dangerous than others, but, even then, the most riskly places aren't exactly neighborhoods, and they are most dangerous during certain hours, if it's "stranger danger" you're considering.

[Rob in CT]

Stranger danger isn't what I'm worrying about, except insofar as fear of this makes the rest of society reluctant to enact policies other than "tough on crime" policies.

I'd say there may be little awareness of the very notion of the social contract

Ok, sure. Ignorance of it, rejection of it... amounts to the same thing, IMO.

But, AFAICT, high-crime neighborhoods are places where lots of people have guns and lots of people have beefs with each other, mostly stemming from turf wars or other disputes related to the drug trade. They mostly kill each other, but not-at-all rarely innocent bystanders caught in the crossfire as well.

That was my impression as well. You could consider this a substitute social contract of sorts. Basically, a "failed state" within a state. The rules the rest of us live by are ignored/not applicable; therefore, gangs.

[hairshirthedonist]

Stranger danger isn't what I'm worrying about,

Sorry. I didn't mean to imply that you were. That was an aside regarding a notion that's generally out there regarding high-crime neighborhoods, at least it seems so from my experience. The calculus seems to be something along the lines of (if not exactly) "the per-capita violent-crime rate in this place is X times the national average, therefore I'm X times as likely to be a victim if I go there." But it doesn't work that way. The macro isn't simply the micro, writ large.

[Countme-In]

What a bunch of gloomy gusses!

Why, in 2010 only 4.8 out of every 100,000 people in the United States were murdered.

Viewed in the warm light of American optimism, that's fully 99,995.2 red-blooded citizens walking the streets unscathed.

And there's more where they came from.

You never hear about them from our sensationalized media, not to mention the green eye shade types in the government keeping track of these negligible numbers for want of more productive activities.

Now, I suppose you could say that 4.8 murder victims annually, added up over the years and decades, will cumulatively begin to whittle down the 100,000 pretty good over time.

But that's the other 100,000, where the murders occur, not the 100,000 I belong to.

Yet we moan and complain and worry and winge.

You don't hear that lucky 0.2 percent of a surviving person complaining, do ya now? No, sirree, he thanks his lucky stars every day when he has the 20% of himself still intact gathered from bed and taken to breakfast to dine through a catheter with the other 0.2 percenters.

True, with only 20 percent of himself to spare, there's not much room to conceal his weapon.

But he also makes a smaller target going forward, which is consolation and then some.

I'll tell you who keeps this hullabaloo going -- the gun manufacturers, that's who.

And the current President, who keeps appearing in public showing empathy for gun victims while being over-melanined.

The ads write themselves.

[russell]

Context is everything:

Top 15 causes of death in the US, by county.

If you don't want to die via homicide, stay the hell out of the southwest and southeast.

Maybe it's the hot weather that makes folks want to off each other.

[hairshirthedonist]

Not too many counties can beat Philly or Baltimore. But the city of Philadelphia is coterminous with Philadelphia County, and Baltimore is an independant city, not included in Baltimore County, or any other county, for that matter, so it gets its own stats. Not exactly apples to apples there.

In any case, the disparity, if you consider for instance the ratios of high counties to the average for a given category, homicide takes the cake (as you'd probably expect).

I live in a county that's at the top for cancer, kidney disease and blood poisoning according to the maps. But those are all still well short of being twice the national average. Baltimore City's homicide rate is more than 6 times the national average.

It's a crazy world. Someone ought to sell tickets. I'd buy one.

[Rob in CT]

http://www.worldlifeexpectancy.com/usa/connecticut-homicide

Tolland County, baby! Woohoo! Represent!

That's what happens when you don't have any real cities... ;)

[Brett Bellmore]

"And, it's always possible that we are, as a culture and a society, simply more violent than other people are. Or, at least, more tolerant of violence, and more willing to accept it as a legitimate way to resolve conflict."

My point would be that we are not "a" culture. We are many cultures. Some of them violent, some of them not. Some of them gun owning, some of them not. There's a big problem here, IMO, with attacking one culture based on crime committed in a different culture.

Gang bangers own guns. NRA members own guns. This does not make gang bangers and NRA members the same culture.

[hairshirthedonist]

There's a big problem here, IMO, with attacking one culture based on crime committed in a different culture.

You seem to be reading something into what you've quoted (or into something else?), Brett.

Anyway, the US does have a cultural composition, even if it's not homogenous. I think that's what russell means by our culture. And it's different from the cultural composition of other countries, however more or less homogenous they may be.

[russell]

My point would be that we are not "a" culture. We are many cultures.

I certainly agree with this.

If you feel it would be more accurate, I'll amend my statement upthread to "we, as a nation, include cultures and societies that are more violent than those found elsewhere".

Because yes, there are large demographic groups within the US that are not prone to violence at all.

This does not make gang bangers and NRA members the same culture.

True again. However it's not gang-bangers who are bumping up the homicide rates in, frex, rural Arkansas.

And don't assume that I think it's NRA members. Because "NRA members" aren't really a monolithic demographic either.

Americans shoot each other much more often than folks in other countries. Especially if you consider OECD countries - countries that are our peers in terms of overall economic and political development - we are 2nd out of 33 in rate of intentional homicide. Only Mexico is worse. Turkey's after us, then it's all kind of noise.

Some of this is no doubt due to the sheer number of guns in circulation in the US, but that is far from the only factor. It's likely not the most significant factor. Other countries with very very low rates of homicide have fairly high rates of gun ownership. So, Switzerland, the Scandinavian countries, Canada -- all countries with low homicide rates -- all have rates of gun ownership well above average.

Americans are, straight up, more prone to acts of violence that their peers in otherwise similar countries. I don't know why that is, but gang-bangers don't begin to explain it. There aren't enough of them, and high rates of homicide are quite common in areas -- lots and lots and lots of areas -- where there are, basically, zero gang-bangers.

So, you tell me why that is. If I were to guess I'd say it's a combination of poverty, lack of basic services, and limited chances for social or economic mobility - basically, Rob in CT's thesis. But I'm just pulling that out of my butt, I really don't know.

[hairshirthedonist]

However it's not gang-bangers who are bumping up the homicide rates in, frex, rural Arkansas.

That's why you have to look at homicides per acre. ;)

[Phil]

We are many cultures. Some of them violent, some of them not. Some of them gun owning, some of them not. There's a big problem here, IMO, with attacking one culture based on crime committed in a different culture.

The dude pictured below walked into a Cracker Barrel restaurant in Cleveland last night with a rifle, killed his estranged wife and his 10-year-old daughter (on her birthday) and sent her sister to the hospital in critical condition, before being shot and killed by police. What culture was he a part of, I wonder?

I bet he was a "gang-banger."

[hairshirthedonist]

I'm going to go out and shoot somebody if I misspell any more words today. (I'm neither an NRA member, nor a gang-banger.)

[Countme-In]

Here's some kind of sick sh*t:

http://www.buzzfeed.com/rosiegray/rick-santorum-signed-up-3-year-old-daughter-for-a

Maybe she's doing target practice from her hospital bed. Daddy, could you reload for me?

The hospital allows heavy weaponry for the nursery in this sick fucking piece of shit Republican murderous culture that al Qaeda didn't have the guts to terminate fully on 9/11 but, hurry, let's get the condoms and birth control pills out of the little wretches' reach, because that might violate an effing encyclical.

Maybe she'll try to shoot the Obamacare doctor when she needs it, out of principle, when her insurance is canceled, since she's well on her way to going over the limit.

You know, I wonder why the NRA and the Small Dick Owners of America don't market to the Gangbangers.

They seem a natural constiuency.

A market. You know, American, in the true murderous sense of the title.

Unless the current arrangement is just a profitable codependent grift of Wayne LaPierre's, the unAmerican murderous vermin, making sure the gangbangers have just enough access to keep the Republican base buying the guns and the ammo and the clips, fearing the over-melatined coming through the suburban bay window.

[wanderingmoderate]

FWIW, the other crime statistic that is considered highly accurate is automobile theft. It is particularly likely to be reported, if for no other reason than insurance compensation would require a police report. I once read (can't remember the source, sorry) that surveys examining crime victimization found that more than 90% of car thefts were reported. In fact, I have a suspicion that if could find the statistics for car thefts not reported, and reported car theft that might be more accurately described as insurance fraud, the two might come close to cancelling each other out. Or perhaps not.
Have to run, may have more thoughts later.

[Turbulence]

I'm working through the ICPSR data and it is...really annoying. Because it was compiled by people who are obsessed with the notion that all data must be placed in ONE AND ONLY ONE TABLE, it seems a bit ambiguous. There are multiple columns for multiple victims and multiple columns for multiple offenders, but no way to link offenders to victims. Moreover, the relationship between offender and victim is a property of the offender.

So, for records with multiple victims but one offender, what does the relationship refer to? The first victim? Maybe. What about the others? For records with multiple offenders and single victims, there's also confusion. I'm looking at one where two men killed one woman who was listed as the girlfriend of both. And of course, multiple offender multiple victim records are highly ambiguous too.

Is it really that hard to have three tables? One for victims, one for offenders, and one mapping the two?

[lise]

What about the issue of 'juking the stats'? (Like in The Wire) I've been reading the NYPD corruption case from Village Voice and it appears to be standard practice in departments using ComStat and doubtful unique to NYC.

http://www.villagevoice.com/2012-03-07/news/the-nypd-tapes-confirmed/

http://blogs.villagevoice.com/runninscared/2012/03/nypd_crime_stat_1.php