What's the best way of comparing violent crime between countries, as I did yesterday? Lots of people are asking me this—in Twitter's usual polite fashion—so maybe you'd be interested in a brief primer.
The short answer is that it's really hard. The two biggest problems are both ones of measurement:
- Different countries put different sets of crimes into the "violent crime" basket.
- Reliability of reporting data differs considerably between countries.
In the US, there are four violent crimes: murder, rape, aggravated assault, and robbery. That's what I used in yesterday's chart. I didn't use a violent crime index from each country, I used just the rates for those four crimes and then converted them myself into a standard form (incidents per 100,000 population).
This takes care of the first problem but not the second, which itself can be broken into two primary components:
- Some crimes simply have vague definitions and therefore leave a lot of discretion in the hands of local police.
- Some countries have better reporting rates and better statistical services that make them publicly available. These are, unsurprisingly, the richer countries.
There's not a lot to be done about #1, and it's a problem even within the United States. The main offender is aggravated assault, which is by far the biggest component of overall violent crime. It's generally defined as either assault with a deadly weapon or the infliction of "severe" bodily injury. This is obviously a judgment call, and we just have to take what we can get.
Problem #2 is more serious. Not only is it likely to produce bigger discrepancies than #1, but it's impossible to get a handle on. In yesterday's chart, for example, Guatemala clocked in with a very low violent crime rate. That's ridiculous! But how do you know it's ridiculous? Only because you already have a preconceived notion of what the "real" crime rate is in Guatemala. But is your notion accurate? How do you know? Is it just because you've heard anecdotal reports, or because you really, truly have some knowledge about crime in Guatemala?
Now, I'm inclined to agree about Guatemala. Its violent crime rate, as calculated by outside sources, marks it very firmly as a high-crime country. However, the officially reported figures are very low.
There's not really a good answer to this problem of unreliable reporting. In yesterday's chart I was performing a comparison with another chart, so I had to use the same countries as the original. That meant including Guatemala. Normally I'd limit comparisons to similar-ish rich countries, where I trust that crime reporting is reasonably accurate. For example:
Now we can finally get to the meat of this post: If comparisons of violent crime are so difficult, should we just use murder rates instead?
The advantage of doing this is that murder is very well defined: If someone is dead and it's not a suicide, then it's murder. What's more, murders tend to get reported to police. It's hard for a dead body in the street to be ignored.
The downside is twofold. First, you're still getting numbers from a national authority, and there's no telling how reliable they are. Second, murder isn't a great proxy for overall violent crime. Partly this is because murder is rare: The US is a high-murder country, but even here it makes up barely more than 1% of all violent crimes. In other countries it's even less. Going solely by murder rates, for example, the US is 5x more dangerous than Germany. Going by overall violent crime, it's about 1.5x more dangerous.
Which do you think is more accurate? I'd go with the violent crime measure in a heartbeat. Thanks to the widespread availability of guns, our murder rate is relatively high, but no one thinks our overall crime rate is 5x that of Germany.
Here's my advice:
- If you're doing comparisons among rich countries, use violent crime. But even though it's a pain in the ass, make sure you use the same set of crimes for every country. It's more reliable than the murder rate.
- If you're making comparisons of relatively poorer countries, murder might be a decent proxy for overall violence. It's not great, but the violent crime rate is often useless, so it's better than nothing.
Finally, just to give you an idea of what different crime rates are like, here is violent crime in the US broken into its four constituents:
Robbery is down 70% while murder is down only 30%. Using only one or the other as a proxy for all of violent crime provides a very distorted picture. This is why the overall rate of violent crime is best if you have even modest trust in the reliability of a country's national crime reporting.
I'd say murder is more reliable.
Or did you mean, "Data on murder vs. data on violent crime: Which is more reliable?"
Actually, some of us are perfectly willing to believe that the overall crime rate in the US is five times as high as in Germany.
Also, if you're counting "murders" as opposed to "homicides", you're introducing a distortion. Many homicides are accounted as justifiable in the US that would be murder in most other countries. An accurate comparison would count all intentional killings, justified or not.
This. All the people shot under stand your ground laws or castle doctrine or whatever would all be considered homicides in many other countries. The US throws out a lot of “justifiable” gun deaths from its murder stats.
In 2019, the U.S. had a reported 16,425 cases of murder and nonnegligent manslaughter (13,927 of these were murder). Excluded from this count are 340 justified homicides by law enforcement and 386 justified homicides by private citizens. This suggests that justified homicides are rare enough that differences in what counts as a justified homicide are not going to have much effect on the reported murder rate.
As should be the case, if you're reporting crimes. Why would you include non-crimes in criminal statistics? I assume you're not suggesting someone should be charged with murder if they shoot someone breaking into their home.
Yes, I am. Trespassing should not be a capital offence.
It's not a capital offense.
I think the intention of Kevin's charts is to report crimes. Why would you want to include justified killings, which are not crimes?
Obviously, trying to compare crime rates when different countries have different definitions of crime is fraught with error.
no one thinks our overall crime rate is 5x that of Germany
Uh. I think it’s closer to 5x than it is to 1.5x. I’ve never felt unsafe anywhere in a German city comparable to feeling unsafe in many US cities or specific neighborhoods in those cities. And I’ve never felt there were lots of guns lurking anywhere around me in Germany, but have felt that way in plenty of places in the US, both urban “bad” neighborhoods and out in rural “god- fearing, good Christian folk” areas, which doesn’t make me feel any more safe. There’s no way German violent crime is just 1.5x that of the US.
The idea that "aggravated assault", which is apparently most of the index, has an internationally-comparable definition as "either assault with a deadly weapon or the infliction of severe bodily injury" is obtuse, and with only a little searching easily falsified. It undermines the credibility of this entire exercise to take that as a basis.
It is always good to set out your standards in advance and stick to them, but when the results don't pass the sniff test at the very least some introspection is called for. The graph even only considering the "rich" countries is absurd (if you put South Africa in that bucket, which I probably would, it's very transparently so - neither Belgium nor New Zealand is in the neighborhood in any meaningful sense, so clearly something is amiss). It's certainly less accurate than the murder one, which has its own problems as noted.
That 6 year old kid who shot his teacher is a little psychopath. Execute the freak now.
https://www.theguardian.com/us-news/2023/feb/07/virginia-teacher-student-shooting-violent-behavior
Is this “violent crime”?
https://nypost.com/2023/02/08/florida-teen-who-attacked-9-year-old-girl-on-school-bus-charged-with-battery/
The 15-year-old Florida boy who was filmed viciously beating a 9-year-old girl on a school bus has been charged with battery, officials announced on Tuesday.
The sickening, widely-shared video shows the much larger teen and another boy mercilessly pummeling the third-grader, a student at Coconut Palm K-9 Academy in Homestead, last Wednesday as she helplessly tries to fend them off on a crowded bus.
Another psychopath. What causes this? What is mis-wired in the brains of these two kids? Poisoned by lead? Abused by adults? It doesn’t matter to me anymore. I’m all out of liberal compassion for the little freaks.
What’s the point of all these posts about crime? We should be happy it’s less than before or somewhere else? Sorry. There is no such thing as a violent crime which is acceptable. Stop making excuses.
Very nice analysis, as usual. If only the NY Times could be as careful and thorough.
Let's put it this way, when in Guatemala don't be surprised when your bus gets stopped at a checkpoint where heavily armed police hide behind sandbags and point machine guns at you while the bus is searched.
I got to say, building the graph with unreliable data does more harm than good. The whole premise of the original article is based on the impression the chart gives. I much prefer Kevin's approach of including countries generally accepted to have good statistics in a particular topic. The original chart also had seemingly arbitrary divisions for each category unless I missed something. I'm short, I would as the writer to redo the chart and then rewrite the article.
Even if you compare by similarly rich countries - excluding Guatemala - you still get that Belgium or France is twice as dangerous as the US. That's ridiculous. I've lived in France, both in small cities and Paris, and the crime rate is not as high as in the US. Your chances of being robbed at gunpoint in France are zero. If there is crime in Paris, it's going to be some pickpocket in the subway.
I’ll agree that your chances of being robbed at gunpoint is much, much higher in the US, but maybe some of you should get more perspective. There is plenty of crime in London or Paris. People get mugged and assaulted there every day. Moreover, that 50% reduction in robbery in the US is very real.
It is true that when I go into “The Big City” there are neighborhoods I avoid, and I am always aware that someone could pull a gun on me at any time. But even though I’ve walked in several different neighborhoods after dark, no one has ever threatened me and I’ve never had to drive through a shootout. Basically, if you don’t try and buy drugs off the street and you never pick fights with strangers your chances of being a victim are pretty low.
And it is once again very nice to see that no one fed the troll.
That's true in Belgium as well. So why is Belgium having twice as much violent crime? France/Belgium don't have the chance of being mugged at gun point, so there's a whole category of violent crime that doesn't exist there. What is the violent crime there that more than makes up for it? This data doesn't make any sense to me.
I'm not an expert on crime in Belgium, but I do recall that various terrorist cells had their origin in Belgium. I read that the perpetrators lived in very high crime areas. Lots of articles written in the past decade or two about Arab crime in France. I don’t have a nuanced view of France’s crime problems, but I would not be shocked to find it is higher than many people think.
You don’t need guns to have a lot of violent crime. Back in my day, which I guess is over a hundred years ago, the US and UK had plenty of crime even though most criminals didn’t carry guns. And we managed to rob, beat, and kill each other just fine. Darn tootin’!
Where does the data come from? I did a quick Google and found a site comparing New Zealand's crime rates and the USA. Every violent crime category was lower for New Zealand, and usually by a lot.
Violent Assault or Aggravated Assault
Again here (NZ) and the UK if I shake my stick at you and shout at you that is Violent Assault or Aggravated Assault
In America I need to HIT you and cause "actual damage" for it to be "Violent Assault or Aggravated Assault
Rape
Rape depends on the reporting rate - in some countries (America) if you report a rape you have to get past a barrage of male obstruction before it is accepted
In Sweden each time a woman is raped counts as one crime
In some countries you cannot "rape" your wife (some US states??)
Robbery
In the USA that means robbing a PERSON
In the UK most of the time that means burgling an EMPTY house
I will extend that - if you burgle a house and break a window or a lock that is considered to be "violence" - which is where we get the interesting information that 90% of burglaries are in empty houses - but 85% of burglaries involve violence
So the ONLY one of your four crimes that is anywhere near to being "Universal" is Murder
So your "First Problem" is nowhere near fixed by selecting those four
100% believable that its 5x worse
As an Australian who has been to the USA five times there is no way, based on my experience, that violent crime is so much higher here than there. Personally I suspect there are a few confounding points that makes comparison very difficult;
1) there are 18,000 police departments in the US and 8 that handle violent crime in Australia (there are a few other orgs that investigate financial/tax crimes etc). I wouldn’t find it hard to believe that the data coming out of the 18,000 PD’s of wildly varying size and administrative quality in the US is as good as that coming out of the 8 departments in Australia
2) similarly there are only 8 justices systems in Australia and their is a level of professionalism and support for victims that may encourage more reporting
3) I wonder if they define crimes the same way eg does Australian data include break and enters when no one is at home?
4) Australian policing numbers and imprisonment rates are far lower per capita than the US - I would expect them to be closer if crime rates were so much higher here.
5) I would also expect crime to be a more salient political issue here if crimes rates were as high but it just isn’t as as big of a political issue here as it is on the states.
I find violent crime stats very misleading in general. For example my county supposedly has a violent crime rate higher than the national average yet I can't remember the last time I read of a violent crime in either of two local papers I read or via the local NPR station. There is no place I know of in the county where anyone would fear to walk alone at night. Lots of people leave their cars and homes unlocked. My only guess is there must be quite a few domestic incidents that are kept quiet. I don't believe we average a single murder in a year.