This is a great article, but you need to dedicate more time to racial differential in homicide rates. In particular, if US black and Latino murder rates were the same as US white murder rates, then the United States would have similar murder rates as our economic peer nations, even though whites own more guns per capita than black folks or Latinos.
I taught public policy at the university level in the late 1990s. The underlying facts have not changed since then, nor have the public’s ignorance of those facts…
Given improvements in emergency medical care for trauma, homicide rates over time should trend down regardless if attempted murder rates are steady (and assuming homicide attempters don't adapt by using more lethal methods).
What I'm wondering is how much of that fall in the homicide rate from the 1990s to 2015 was due to emergency rooms getting much, much better at saving people's lives, including bullet wounds (and thus, not becoming a statistic). I've heard that much of the decline in homicide rates in recent years is due to this factor alone. I have no way to verify this, but I'm wondering if anyone has looked at the issue and tried to adjust for it in the statistics, or if it's just invisible.
The National Crime Victimization Survey -- which surveys a large random sample and asks whether people have been victims of crime -- shows clearly that there also was a large decrease in non-fatal violence from the '90 on. So it's certainly not the case that it's just people being saved more.
With that said, I certainly do think medical advances must play some role, particularly for long time frames. The time series of homicide rates I plot starts from the early 20th century, a time when not even penicillin was invented. So surely since then medical advances must have some influence.
There are some who have looked into this topic (e.g., Dobson, 2002; https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1124155/). They claim that "Murder rates would be up to five times higher than they are but for medical developments over the past 40 years." While I believe there is some effect, this large of an effect seems implausible to me, particularly when it's not observed in non-fatal violence victimization statistics (which should show a large increase).
Excellent data compilation, thank you for putting all this together.
I am not really sure how to parse this portion: "Many homicides were precipitated by another crime (22.9%), and the crime was in progress in 66.0% of those cases. The most common precipitated crimes were assault/homicide (38.9%),"
What does it mean that a homicide was precipitated by a homicide? Was it a different homicide target and the homicide in question was ancillary to the main target?
They grouped assault and homicide into one category. This means that many homicides were precipitated by an assault (a subset of which are lethal). In other words: a significant fraction of homicides committed are part of a string of multiple assaults occurring close together (and some of the other assaults may or may not be lethal).
Oh, ok, so for example a few guys start beating down some victim and the victim gets killed, but by accident? Or the perp was fighting one guy, then kills the guy's buddy who gets involved?
I guess I am confused because it seems like every homicide would precipitate from assault, unless the victim was entirely unaware of the perpetrator. It kind of feels like if the chart with "Number of Arrests before Incarceration" had a column for zero; how did they get incarcerated if they weren't arrested? I don't think I understand the definitions behind the statistic.
I'm fairly sure that a homicide being "precipitated by an assault" means that it's precipitated by an assault *other than the homicide in question*. A different crime happened before the homicide in question.
So an example: a man attacks one person, then attacks a different person (could be anything from a bar fight, to gang fight, to familicide). If the second person dies, it's a homicide precipitated by an assault. If the first person dies as well, then the latter attack would be a homicide precipitated by a homicide.
Ok, that makes sense. Thank you for clearing that up, that was some tricky data to think about without knowing quite what the data gathering and definitions looked like. I think in my head I was thinking precipitated by meant "Crime X led to the homicide which happened in the process". So like robbery leads to a homicide because the victim struggles and the mugger knifes them, which might not have happened otherwise. Seeing it as less causal and more just "these happened to occur around each other" makes it more clear.
Nice work - more where this came from please! It is refreshing to see such a strong effort to aggregate the numbers. I wish more people who are anxious to share their unearned wisdom were so inclined.
Good overview, but you use homicide rate as the violent crime rate in the US, and there are some studies that say that both the overall crime rate and the violent crime rate minus homicide is lower in the US than in much of Western Europe, or at least had been lower in the 2000s, and that the US homicide is an outlier. Here: https://sci-hub.ru/10.1111/j.1468-0327.2011.00267.x
I don't use homicide rate specifically for USA. When I make international comparison I use homicide rates for all the countries. I explain why I focus on homicide in the introduction, namely the far greater reliability of homicide statistics than for other crimes.
This also leads me to your referenced study: its results simply cannot be trusted. The measure they use (the number of offenses reported to the police) is not a good measure of crime activity, and especially not as a comparison between countries. Justice systems differ a lot between countries and so do the propensity of people to report crimes they experience.
As you say, homicide is the most reliable crime to compare between countries, because homicide leaves bodies that you can't easily undercount, and most of the time it's very obvious if someone was killed. At least among developed countries or even somewhat functional countries, you don't expect them to miss many homicide bodies or misreport many of them.
That said, I think the issue raised by the study I referenced is an interesting one. The authors of that study did another study, talking about the reliability of crime statistics, and in it they note that in the US homicide and other crimes fell at the same time, while in Europe there is a tendency for homicide to stay flat or fall while other crimes increase. Now, it seems somehow unlikely, in my view, that the US saw less crime reporting in the same period in which it saw less homicide. Similarly, it seems a little unlikely that European crime didn't increase, but that reporting just got better and better, because among other things, we know that Europe has imported immigrant groups with high crime rates.
Of course, it is possible, perhaps probable, that the US was reporting much less crime than Europe at the beginning of the series, and that the US continued to have more crime than Europe at the end of the series, despite any crime decline that the US experienced and that was reflected in the statistics.
In the other study, the authors hypothesize that homicide and other crimes follow the same trend in the US because in the US other crimes more easily turn into homicide, due to the greater availability of firearms. Here: https://sci-hub.ru/10.1007/s10657-017-9555-6
Super informative. Thank you!
This is a great article, but you need to dedicate more time to racial differential in homicide rates. In particular, if US black and Latino murder rates were the same as US white murder rates, then the United States would have similar murder rates as our economic peer nations, even though whites own more guns per capita than black folks or Latinos.
https://hwfo.substack.com/p/real-talk-about-race-and-murder-rates
Great summary.
I taught public policy at the university level in the late 1990s. The underlying facts have not changed since then, nor have the public’s ignorance of those facts…
Are parole violations included in that Recidivism chart? 80% getting arrested again is WILD
Given improvements in emergency medical care for trauma, homicide rates over time should trend down regardless if attempted murder rates are steady (and assuming homicide attempters don't adapt by using more lethal methods).
What I'm wondering is how much of that fall in the homicide rate from the 1990s to 2015 was due to emergency rooms getting much, much better at saving people's lives, including bullet wounds (and thus, not becoming a statistic). I've heard that much of the decline in homicide rates in recent years is due to this factor alone. I have no way to verify this, but I'm wondering if anyone has looked at the issue and tried to adjust for it in the statistics, or if it's just invisible.
The National Crime Victimization Survey -- which surveys a large random sample and asks whether people have been victims of crime -- shows clearly that there also was a large decrease in non-fatal violence from the '90 on. So it's certainly not the case that it's just people being saved more.
With that said, I certainly do think medical advances must play some role, particularly for long time frames. The time series of homicide rates I plot starts from the early 20th century, a time when not even penicillin was invented. So surely since then medical advances must have some influence.
There are some who have looked into this topic (e.g., Dobson, 2002; https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1124155/). They claim that "Murder rates would be up to five times higher than they are but for medical developments over the past 40 years." While I believe there is some effect, this large of an effect seems implausible to me, particularly when it's not observed in non-fatal violence victimization statistics (which should show a large increase).
Comparing homicide rates, not by economic tier but by geographic region, has some interesting insights as well.
Excellent data compilation, thank you for putting all this together.
I am not really sure how to parse this portion: "Many homicides were precipitated by another crime (22.9%), and the crime was in progress in 66.0% of those cases. The most common precipitated crimes were assault/homicide (38.9%),"
What does it mean that a homicide was precipitated by a homicide? Was it a different homicide target and the homicide in question was ancillary to the main target?
They grouped assault and homicide into one category. This means that many homicides were precipitated by an assault (a subset of which are lethal). In other words: a significant fraction of homicides committed are part of a string of multiple assaults occurring close together (and some of the other assaults may or may not be lethal).
Oh, ok, so for example a few guys start beating down some victim and the victim gets killed, but by accident? Or the perp was fighting one guy, then kills the guy's buddy who gets involved?
I guess I am confused because it seems like every homicide would precipitate from assault, unless the victim was entirely unaware of the perpetrator. It kind of feels like if the chart with "Number of Arrests before Incarceration" had a column for zero; how did they get incarcerated if they weren't arrested? I don't think I understand the definitions behind the statistic.
I'm fairly sure that a homicide being "precipitated by an assault" means that it's precipitated by an assault *other than the homicide in question*. A different crime happened before the homicide in question.
So an example: a man attacks one person, then attacks a different person (could be anything from a bar fight, to gang fight, to familicide). If the second person dies, it's a homicide precipitated by an assault. If the first person dies as well, then the latter attack would be a homicide precipitated by a homicide.
Ok, that makes sense. Thank you for clearing that up, that was some tricky data to think about without knowing quite what the data gathering and definitions looked like. I think in my head I was thinking precipitated by meant "Crime X led to the homicide which happened in the process". So like robbery leads to a homicide because the victim struggles and the mugger knifes them, which might not have happened otherwise. Seeing it as less causal and more just "these happened to occur around each other" makes it more clear.
Thanks again!
Nice work - more where this came from please! It is refreshing to see such a strong effort to aggregate the numbers. I wish more people who are anxious to share their unearned wisdom were so inclined.
Good overview, but you use homicide rate as the violent crime rate in the US, and there are some studies that say that both the overall crime rate and the violent crime rate minus homicide is lower in the US than in much of Western Europe, or at least had been lower in the 2000s, and that the US homicide is an outlier. Here: https://sci-hub.ru/10.1111/j.1468-0327.2011.00267.x
I don't use homicide rate specifically for USA. When I make international comparison I use homicide rates for all the countries. I explain why I focus on homicide in the introduction, namely the far greater reliability of homicide statistics than for other crimes.
This also leads me to your referenced study: its results simply cannot be trusted. The measure they use (the number of offenses reported to the police) is not a good measure of crime activity, and especially not as a comparison between countries. Justice systems differ a lot between countries and so do the propensity of people to report crimes they experience.
As you say, homicide is the most reliable crime to compare between countries, because homicide leaves bodies that you can't easily undercount, and most of the time it's very obvious if someone was killed. At least among developed countries or even somewhat functional countries, you don't expect them to miss many homicide bodies or misreport many of them.
That said, I think the issue raised by the study I referenced is an interesting one. The authors of that study did another study, talking about the reliability of crime statistics, and in it they note that in the US homicide and other crimes fell at the same time, while in Europe there is a tendency for homicide to stay flat or fall while other crimes increase. Now, it seems somehow unlikely, in my view, that the US saw less crime reporting in the same period in which it saw less homicide. Similarly, it seems a little unlikely that European crime didn't increase, but that reporting just got better and better, because among other things, we know that Europe has imported immigrant groups with high crime rates.
Of course, it is possible, perhaps probable, that the US was reporting much less crime than Europe at the beginning of the series, and that the US continued to have more crime than Europe at the end of the series, despite any crime decline that the US experienced and that was reflected in the statistics.
In the other study, the authors hypothesize that homicide and other crimes follow the same trend in the US because in the US other crimes more easily turn into homicide, due to the greater availability of firearms. Here: https://sci-hub.ru/10.1007/s10657-017-9555-6
The answer people don't want, America has been the dumping ground for undesirables, is obvious.