Thursday, March 28, 2013
Comparing England (or UK) murder rates with the US: More complex than you thought
Thoughts on the Gun Debate part 4a — wherein I fix my oopsie in regard to comparing US and English murder rates... Oooops. It turns out that I was wrong.
This post edited after comments were reviewed on 3/29/2013. Edits from that date will be noted with an Edit tag.
First, lets get this out of the way. I'm not going to fuss here about the terms "England", "The U.K", "Great Britain", "The Olde Country" or "The folks with those classy accents who live on some islands off the coast of Europe." I'll be careful to cite things in the links, but don't chew me up on where I'm talking about.
On to the oopsie. I have frequently in this series referred to the English murder rates as historically low and currently very low compared to US murder rates. I blandly accepted the murder statistics published by the UK Home Office as definitive. I overlooked the details of what and how the English counted "murders." It turns out that was a big mistake. (I was first turned onto my error by this post at Extrano's Alley.)
I fell into a definitions trap you may not be aware of. The shortest version is this. We count and report crimes based on initial data. The Brits count and report crimes based on the outcome of the investigation and trial. Yep, that says what I meant it to say.
In the US, the count of people murdered kept by the FBI is pretty darned straightforward. Got a body, not natural causes, not suicide? Must be murder of one sort or another. Count it.
So, if you ask the FBI, they will tell you that for 2011 there were 14,022 murders or non-negligent manslaughters. On the same line of that chart, they tell us the population was 292,364,075 which gives us a "murder" rate of 4.8 per 100,000 population. Those counts are based on crimes reported by local police agencies. They say nothing about the clearance rate, nor if anyone was ever identified or charged or convicted or whatever. Body, not natural, not negligent, homicide. Duh.
Now, on to England. It turns out that the Home office is very restrictive in what they report as "murders." Still, looking at the detailed report for 2010/2011 the Home office tells us that in the reporting period there were 636 murders "provisionally recorded" for a murder rate of 1.15 per 100,000 --- less than 1/3 the murder rate in the US. (See page 16 of the source document)
I've reported these numbers blindly many times, and quoted sources with many (sometimes silly) explanations for the lower murder rate in the UK. There's a problem with that as it turns out. What about all those murders which were not solved? The ones where a conviction wasn't gotten? The ones where the appeals are still on-going? Not only that, but when exactly were these homicides performed? The nice folks at the Home Office tell us:
Homicides are often complex and it can take time for cases to pass through the criminal justice system (CJS). Due to this, the percentage of homicides recorded in 2010/11 (and, to a lesser extent, thoserecorded in earlier years) to have concluded at Crown Court is likely to show an increase when thenext figures from the Homicide Index are published in 12 months‟ time.
But in any event, according to a report to a select committee of Parliment:
Since 1967, homicide figures for England and Wales have been adjusted to exclude any cases which do not result in conviction
Note that the numbers provided were for murders "recorded" in 2010/2011, not murders "performed" in 2010/2011. The killing might have happened a decade ago. As a matter of fact, when a serial murderer was found out and convicted of some 172 odd killings over the course of two decades, all his murders got counted onto one year! Quoting from the previously cited Home Office report:
Caution is needed when looking at longer-term homicide trend figures, primarily because they are based on the year in which offences are recorded by the police rather than the year in which the incidents took place. For example, the 172 homicides attributed to Dr Harold Shipman as a result of Dame Janet Smith‟s inquiry took place over a long period of time but were all recorded by the police during 2002/03. Also, where several people are killed by the same principal suspect, the number of homicides counted is the total number of persons killed rather than the number of incidents. For example, the victims of the Cumbrian shootings on 2 June 2010 are counted as 12 homicides rather than one incident in the 2010/11 data. (Homicides, Firearm Offences and Intimate Violence 2010/11 page 16)
OOOoooooops. We're not comparing apples to apples, we're comparing apples to meatloaf.
Wait! I have a clever idea! Instead of going to the Home Office crime stats, let us go grab the death register numbers! (Link to excel spreadsheet) (essentially the same as the US CDC wisqars data) That leads to some odd things... (See Table 2 in the linked spreadsheet.)
In 2011 329 people died from "assault", 27 by poisoning (not suicide or work related), 361 by strangulation (not suicide), 127 by non-accidental or suicidal drowning, 7 with guns, 2 with explosives, 20 by stabbing, 62 pushed from a high place, 21 run over, and another 198 of "other specified events in various places" .
I make that 1154 violent deaths of interest to the police which would in the US be reported as murders, and that doesn't include every death that might be a murder since the "cause of death" of a murder or manslaughter victim might well be an infection or other medical complication resulting from an injury during a crime or assault. That death would be classed as murder in the US but I can't pull it out from the causes of death numbers. Never the less at a minimum this gives us about double the number of "Murders" in England as was reported in the Home Office crime stats. Edit Quick math says that if 636 deaths was a rate of 1.15 per 100,000 then 1154 is 2.08 per 100k. Still substantially lower than the US rate, but substantially higher than the Home Office number.
There is another data source. You would think that you could go to the UK's dept of Justice and look at the outcome of Coroner's inquests to find homicides. Yep, you'd think that, but over the last decade coroners have taken to producing "narrative" verdicts instead of calling something a homicide Edit or an accident or negligent or whatever. It's a way of avoiding making a causal decision at the level of a cornoner's inquest. This is due to two different forces acting on them. First, the 1988 Coroner's act. Quoting from the decision in the "Middleton" case:
The 1988 Act recognises that a death which is the subject of an inquest may also be the subject of criminal proceedings, and also recognises the general undesirability of investigating publicly at an inquest evidence pertinent to a forthcoming criminal trial. In a departure from previous practice, section 11(6) of the Act provides:
"At a coroner's inquest into the death of a person who came by his death by murder, manslaughter or infanticide, the purpose of the proceedings shall not include the finding of any person guilty of the murder, manslaughter or infanticide; and accordingly a coroner's inquisition shall in no case charge a person with any of those offences."
Thus the inquest jury may no longer perform its former role as a grand jury. Section 16 of the Act (and rules 27 and 28 of the Rules) make provision for the adjourning of an inquest when criminal proceedings are or may be pending on certain specified charges or in certain specified circumstances (but not solely because any criminal proceedings arising out of the death of the deceased have been instituted: rule 32 of the Rules). After the conclusion of criminal proceedings the coroner may resume the adjourned inquest "if in his opinion there is sufficient cause to do so" (section 16(3)). Section 17A makes provision for the adjourning of an inquest when a public inquiry into a death is to be conducted or chaired by a judge. A coroner may only resume an inquest so adjourned "if in his opinion there is exceptional reason for doing so", and then subject to conditions (section 17A (4)). (Highlighting above is mine.)
The second force is the Middleton decision itself... which I'll let you read. From an American's perspective, it's very odd.
Sorry about the long aside, back to murders. Remember that 657 number from the Home Office? The Coroners only called 229 of the cases they determined a cause of death on a homicide, and in 4400 cases they filed a "narrative verdict" describing the cause of death in a narrative manner without putting it in a category. If those 4400 cases are what we would normally call murders Big Edit starts here that would suggest that the correct number of "violent deaths of interest to the police" is on the order of 4700 for 2011, then the UK murder rate is 8.5 per 100,000 or about 177% of the US murder rate. Now, honestly, we don't know what conclusion as to cause the coroner would have reached if they weren't using It's entirely possible that very few of them would have been classed as homicides. We don't know. My point here isn't that the English death rates should be quoted from the highest available but rather, no matter which source I attempt to use, I can't actually get an apples to apples comparison. The data simply isn't available.
In the comments, Cam found this interesting link on the effect of narrative verdicts on mortality statistics from 2010. Cam helpfully runs the numbers to suggest that using this percentage breakdown of the narrative verdicts would increase the UK murder count by 308. So, we have the "original 657 plus the "reconstructed" coroner's number of 308 giving 965 potential homicides. This is about 200 less than I got from the cause of death numbers, but I think these numbers are for England only and don't include Wales, where the coroner's system is different. (Lots of countries and lots of legal systems in a very small space. The entirety of England, Scotland, Wales and both Irelands would fit comfortably in Kansas and Nebraska with lots of space left over.) So I think this analysis is confirming.
So, essentially, I conclude that the English murder rate, when analysed using the same criteria as the US murder rate, based on inputs rather than outcomes, is on the close order of double the Home Office rate of 1.15.
In the first version prior to edits of this post I noted that I had English Journalism sources which cited a much higher number of approximately 4700 deaths in 2011 which would be called "Violent deaths of interest to the police" which would generate a murder rate of nearly 8.5 per 100,000. Sadly, my original sources have disappeared into the link-rot common on commercial web sites. I am searching for, and if I can find printed copies of those sources, or alternate sources of the same information, I will put those links into a future Blog post. End of Big Edit
I know, I know, there's no way I'm going to get most of my friends to read this, much less get people like Piers Morgan and his ilk to believe it much less repeat it, but still, here's the short version:
The murder rate in the UK is either equal to or higher than the murder rate in the US. (Sources not available. See reduced conclusion instead.)
The murder rate in the UK according to US standards is double or higher than their reported rate. It may be impossible to produce an actual apples to apples comparison number from official sources. It is not 15% of the US rate.
Note also that this is the MURDER rate, irrespective of the METHOD of the murder. I utterly refuse to get into the discussion of "gun violence." A death is a death.
Closing note: Thanks to the excellent commentators for additional sources and for math error catches. You all are great. -_ Rick
Rick Boatright at 10:54 PM
ChrisMarch 29, 2013 at 11:00 AM
Rick, you note the US murder rate as being 4.8 per 100,000 early on, and at the end you estimate the UK rate as being 4.7 per 100,000 but go on to characterize this as being higher than the US rate. One of those numbers would seem to be wrong.
As a totally unrelated aside, in one of the cases cited in the Middleton decision, it appears that some Italian prisoners can go on "leave" from prison. I'm mildly curious as to how the Italian authorities define that term, and how it is administered. And how often such prisoners participate in crimes such as the murder in the cited case.
Liberty's AdvocateMarch 29, 2013 at 1:38 PM
The World Bank says the UK population in 2011 was 62.641 million.
Compared to the US at
Hope this helps.
perlhaqrMarch 29, 2013 at 1:39 PM
Chris: Yeah, there's some kind of math error somewhere, although I don't know which number is wrong.
Presuming the 1.15 / 100,000 people for 636 murders is correct, backing those numbers out gives a British population of 55,304,348. If the 4,700 "violent deaths of interest to the police" is correct, that gives a "violent deaths of interest to the police" rate of ~8.5 per 100,000 which is way more than 15% higher than 4.8 per 100,000. In fact, that would be more like 77% higher.
So, yeah. Something got jumbled somewhere. Math check!
Aaron S.March 29, 2013 at 2:21 PM
Any word on when we'll see links justifying calling all 4400 "narrative verdict deaths" murders?
Cam.March 29, 2013 at 6:55 PM
This comment has been removed by the author.
Cam.March 29, 2013 at 6:57 PM
(Sorry for the double post — couldn't see an easy way to edit a dumb mistake I made the first time.)
At the bottom of my response, you'll find a links to an interesting report published in Health Statistics Quarterly in 2009, which discusses the deaths that were described as "narrative verdicts". In my brief searching around, I haven't found something similar for 2011 yet, but the 2009 data are interesting because they suggest that the vast majority of these "narrative verdicts" do NOT reflect deaths from homicide/murder.
If you look at figures 1 and 2, you'll see that of the 3012 narrative reports filed by the coroners in 2009, slightly over 60% of them dealt with deaths from "disease-related causes", while slightly less than 40% dealt with deaths from "external causes".
Furthermore, Table 1 indicates that 33.3% of these 3012 narrative reports reflect "accidental" causes of death.
That means that the proportion of these narrative reports that reflect neither "disease-related causes" nor "accidental" causes of death is only about 7% (that is, 100%-(40+33.3); I'm assuming here that there isn't overlap between the disease-related deaths and the accidental deaths, which seems to be what the authors are saying). Even were we to assume, for sake of argument, that all of those were homicides (which is still, I think, an unsafe assumption), that means that the maximum
number of additional deaths that were likely to reflect murder/homicide in 2009 is 210.
As I said, I haven't run down the numbers for 2011, but let's assume (1) that the number of narrative reports dealing neither with disease nor accidental deaths was similar in 2011 to 2009 (7%); and (2) that all of these were deaths that would be categorized as homicide/murder in the united states. That means that at a high estimate
, we're only looking at an increase to the official figures of 308 murders/homicides in 2011, not 4400. That's a difference of a whole order of magnitude. And, of course, it means that the murder rate in the United Kingdom probably is quite a bit lower than it is in the United States.
Here's the link:
Rick BoatrightMarch 29, 2013 at 8:50 PM
Thanks to everyone.
First, yep, math errors. Will fix and note as edited. Thanks.
Second, regarding narrative verdicts, nope, I had not seen that 2010 report, thanks, I'll edit, note and adjust. Note please that I didn't insist that they were all homicide, but rather that I couldn't TELL. However, your 2010 numbers suggest that the 1100 some odd "violent deaths not obviously suicide or accident" taken from the causes of death numbers -are- more comparable to the US murder rates than the 600 some odd cited by the Home Office.
I.E, the rate turns out to be about double what the cops are reporting. Still thanks for that 2010 link, I'll cite and edit the post.
Rick BoatrightMarch 29, 2013 at 9:46 PM
And that's going to be it for editing THIS post. More will hopefully be revealed on a future rock if I can find sources and time.
In the mean time, and for one last time, thanks to everyone for your assistance.
Cam.March 29, 2013 at 9:52 PM
The other potential problem here is that it isn't clear to me which deaths in categories Y10-Y34 in the 2011 data should be characterized as "violent deaths not obviously suicide or accident."
Take the drowning deaths, for instance. The category here is simply labelled "drowning and submersion, undetermined intent." While that may mean that these were suspicious drownings that would have been reported as homicides in the United States, it could just as easily be the case that they were classified in this way because authorities could not determine whether they were accidents, or whether they were suicides.
It's the same with the 62 deaths by falling, which are reported there as "falling, jumping, or pushed from a high place." Again, not all of these necessarily would have been reported as murders in the United States — i.e., it's possible that "foul play" would have been ruled out, but that investigators still could not discriminate between accident and suicide.
Cam.March 30, 2013 at 7:41 AM
Sorry for yet one more post, but this topic interested me enough to get me reading through mortality reports for several hours.
Anyway, as a follow-up to my last point about deaths reported in categories Y10-Y34, here's what the official metadata for the Death Registrations has to say about the way in which deaths are coded (the link to the source is at the bottom):
2.16 Assault and intentional self-harm
Numbers of deaths from assault (homicide in ICD-9)
It is possible to make alternative assessments about the number of deaths that may be
attributed to assault. Different estimates have been used, both in the past and presently,
within any publication and even within a single table of data. The two main estimates used in
the main tables are as follows:
(i) The number coded to X85–Y09. This is the basic ICD classification to which all
assaults should eventually be assigned.
(ii) The number coded to X85–Y09, plus those coded to U50.9 (event awaiting
determination of intent). This takes account of accelerated registrations, most of
which are eventually coded to an assault code (see section 2.15).
Numbers of deaths from intentional self-harm (suicide in ICD-9)
Mortality statistics: Metadata
Office for National Statistics 24
As with assault, it is possible to make two separate estimates of the number of deaths
annually from intentional self-harm:
(i) The number coded to X60–X84. This is the basic ICD classification to which all
definite intentional self-harm verdicts are assigned
(ii) The number coded to X60–X84, plus those coded to Y10–Y34 (event of
undetermined intent). This takes account of most deaths where an open inquest
verdict was returned, but excludes all deaths that are pending investigation.
So, in other words, the ONS thinks that the deaths that are the result of an "event of undetermined intent" in categories Y10-Y34 are more likely to be suicides than homicides, and are not deaths that necessarily sparked criminal investigations as homicides.
The method they recommend for estimating homicides, when applied to the 2011 data, gives a total of 699 deaths (329 in the "assault" category, X85-Y09, and 370 in the "inquest adjourned" category, U50.9) That's actually pretty close to the 636 homicides reported by the Home Office.
Link to metadata:
UnknownMarch 30, 2013 at 8:38 AM
Is there no way to determine how many convictions there were for murder in the United States?
If so, I think that gives you the closest 'apples to apples' comparison.
According to this slightly dated chart, that lists murder convictions per 1,000 people at .01 for England, and .06 for the United States. I don't know necessarily what this means though, as I'm unable to control for the effectiveness of the justice system, plea bargains, etc.
 - http://bjs.gov/content/pub/html/cjusew96/cpp.cfm
GregMarch 31, 2013 at 7:14 PM
I wonder if there is something similar with canada's stats?
mollyMarch 31, 2013 at 11:05 PM
wow...if we counted only 'cleared; r convicted murder cases..the number would be soo very low!!
pdxr13April 1, 2013 at 8:30 AM
This article is more confirmation of advice given by Grandparents:
"Avoid People doing stupid things, in disreputable places, at unlikely times, with contraband, and you will have a long and happy life"
Watch your step. Wear light body armor while riding public transit or in a car smaller than a Ford Taurus. Get a permit and training, then carry a firearm everywhere legal to do so. Wear a seatbelt in cars, helmet on bike, and leathers on motorcycle. Become a better swimmer.
Lies, Damned Lies, and Statistics.
Dean WeingartenApril 1, 2013 at 5:26 PM
European Murder Rates Compared to the United States: Demographics vs Guns
Monitoring people's right to effective self-defence..A well regulated militia, being necessary to the security of a free state, the right of the people to keep and bear arms, shall not be infringed
HOW ODD THAT MASSACRES MOSTLY HAPPEN IN "GUN-FREE ZONES"! When will the brain-dead Left wake up and draw the obvious conclusion? Gun bans kill kids
Murder rates are determined by culture, not availability of weapons.
Alan WadeApril 3, 2013 at 12:04 AM
Best UK VPN circumvents these restrictions and enables anyone living outside UK to access every website on the internet.