This is a good analysis, but I guess I'm left wondering whether the NCVS is reliable, more than anything. Murder, of course, is only a small % of violent crime (<1), but there blacks are victimized at a much higher rate than their share of the population (about 4x). Is it reasonable that they are the victims of non-fatal violence at a lower rate than their population and a much lower rate than their murder? This seems to strain credulity.
According to the NCVS, black people are actually slightly more at risk of violent criminal victimization than white people (so it's not true that they're underrepresented among victims of non-lethal violent crime), but only 4% more so I think you make a good point because this seems hard to believe given how overrepresented they are among victims of homicide. I suspect that the most at-risk black people are undersampled due to racially biased non-response rates.
Robbery is extremely unusual in it's victim and offender demographics. Classify it as a crime against property, as the NIBRS now does, and you get a completely different picture of crimes against persons. In general it's a mistake to consider this issue without disaggregating by crime. Details here:
This paper looks interesting, but according to the NCVS, robbery makes up only 10% of violent crime, so I'm not sure how it could make such a large difference if we excluded it and I also don't think it excluding it from the data on violent crime is legitimate if we're trying to explain racial/ethnic patterns of victimization. I agree that it would be interesting to disaggregate by type of crime though, but I'd be more interested in what things look like once you exclude domestic violence, which is going to be overwhelmingly intraracial by its nature.
If you just look at what NIBRS now classify as crimes against persons (Assault, Homicide, Human Trafficking, Kidnapping, and Sex Offenses) I'd be interested to see the results. You probably also want to exclude simple assault, just look at aggravated assault. Robbery is now grouped as a crime against property. Grouping by motive makes a lot of sense for some purposes, for reasons discussed here:
A lot of discourse around this issue is implicitly (and sometimes explicitly) essentialist, there's no contemplation of the possibility that behavior is driven by incentives. Back in the 1890s, the Mississippi Supreme Court considered robbery and murder to be "robust crimes of the whites" so they did not strip convicted robbers and murderers of the right to vote. Instead they disenfranchised those convicted of fraud and other "furtive" offenses which they associated with Black Americans. Now the stereotypes have been reversed, but not because groups of people have magically changed character. The case is discussed in the book linked above if you're interested.
If I want to disaggregate, I'll have to use the microdata, but I don't understand the logic for excluding robbery. I understand that you made the case for grouping by motives in your book, but I don't have time to read it at the moment, so I was hoping that you could explain the logic succinctly here.
In any case, if I do take a look at the microdata to disaggregate, I'll publish some kind of update and will upload my code so anyone can easily group crimes in any way they want and check how that affects the results of my analysis.
Again, personally I'm inclined to think that people who commit robberies are similar to people who commit aggravated assault and I'm more curious to see how excluding domestic violence would affects the results, but I could be wrong.
Robbery is special beacuse it is a crime of appropriation in which victim and offender interact, so stereotypes come into play. Can't happen with burglary, larceny, or motor vehicle theft in most cases. So the victim and offender demographics are completely different. Lots of data on this based on NCVS, arrests, etc in the book, and some in the article in press. The point is this, if you want to talk about interracial crime and you really want to understand it rather then just reinforce an essentialist narrative, you have to disaggregate. Demographics of murder, kidnapping, sex crimes, aggravated assault, and arson are all very different from robbery. There's a chapter on robbery in the book, pdf available free in most academic libraries.
Indeed. Buried in this is something that seems kinda obvious: You're going to rob someone who's more likely to have more money. That's going interact with race. I didn't actually read this post in detail so I don't know if that was taken into account but I saw your name in the comments and i know you're worth reading.
I heard you on Glenn's pod, (BTW, interesting discussion on using betting markets as a focusing mechanism in voting prediction) and found your quick mention of geographic location as a confounder on these issues. Do you have a link?
My own anecdotal experience is that property crime is vastly underreported, especially in bad crime areas. The first time somebody broke into my car in Baltimore I reported it. The cops said there was a 0% chance they would investigate and that basically they would only do anything if it was murder. I never reported any of the other break ins.
Indeed. Buried in this is something that seems kinda obvious. You're going to rob someone who's more likely to have more money. That's going interact with race. I didn't actually read this post in detail so I don't know if it was taken into account but I saw your name in the comments and i know you're worth reading.
I heard you on Glenn's pod, (BTW, interesting discussion on using betting markets as a focusing mechanism in voting prediction) and found your quick mention of geographic location as a confounder on these issues. Do you have a link?
More money *in their possession* and this is correlated with race but not in the way you might expect; resistance is by far the more important concern, there is good evidence for this; resistance is high when you have more money in your possession relative to total wealth; links below for more on this and the geography of lethal force:
As I said before, I don't think your argument about why robberies should be excluded very convincing, but in any case it doesn't matter. After we had this conversation a few months ago, I redid my analysis using the microdata to exclude robberies and it didn't affect my argument, i. e. there are still large differences between the observed distribution of victims by offender groups and the expected distribution under the third model I consider. Obviously, the various caveats I make at the time on how to interpret this result still apply, but excluding robberies doesn't fundamentally change anything. I wanted to look at a few other things, but I didn't have time, so I didn't write a follow-up on this. But I'll try to do so eventually and I'll upload the code that allows for disaggregation by type of crime using the NCVS microdata.
Fryer is not so easy to classify. His piece on DEI for example is quite nuanced and he pushed back on her blanket statements in the video. Bari herself I find to be quite reasonable on many issues and have been a guest on her podcast once. Links below.
I'll watch the podcast but my assessment of Bari is quite different from yours, particularly with her long-standing hypocrisy on "free speech" with respect to criticism of Isresl.
As for Fryer, while I can certainly understand his comments about karma here:
I would like to think he's rise above that kind of response and certainly his lack of response to Bari's implying that Gay endorsed "calling for the genocide of Jews" was really not even understandable.
Yes I agree about Bari on Israel, that is one issue where she won't countenance dissent, unlike Noam Dworman for example. Would love to see Norman Finkelstein or Rashid Khalidi on FP but not likely, both have been on Live from the Table.
Roland is a part time stand-up so I think he was just performing a bit but I see your point.
"The point that it was making is that, if you listen to the media, you’d think that black people are much more at risk of being victimized by white people than the other way around, but in fact the opposite is true."
I've never seen anyone in the media claim that for crime overall, blacks are much more at risk of being [violently] victimized by whites than vice versa. Is there any research illustrating that the media frequently makes this claim?
(I have seen people in the media say that blacks are more likely to be victimized by white cops, but that's not a generalization about all violent crime.)
Anecdotally, a lot of local news reporting is dedicated to crime, and it's pretty common to see Black faces when they put up mug shots of offenders. I can't speak to percentages, but the media is certainly willing to report on violent criminals who are Black.
I don't have a Twitter account, but I saw your more elaborate bar graphs for rape, robbery, aggravated assault and simple assault. Surely it can't be right that neither whites nor Asians ever rob Asians?
Interviews with criminals may be relevant. In Wright and Decker’s Armed Robbers in Action, they interview a few dozen (mostly black) active robbers in St. Louis (snowball sampled through a reformed ex-robber). Robbers consider several factors when deciding who to rob, and there is some disagreement among them on who is the best target. But there was general agreement that whites were safer targets — less likely to fight back. Given the demographics of local neighborhoods race probably is a decent indicator of whether someone has experience with violence or follows the code of the streets with its ethic of retaliation and toughness.
This is a good analysis, but I guess I'm left wondering whether the NCVS is reliable, more than anything. Murder, of course, is only a small % of violent crime (<1), but there blacks are victimized at a much higher rate than their share of the population (about 4x). Is it reasonable that they are the victims of non-fatal violence at a lower rate than their population and a much lower rate than their murder? This seems to strain credulity.
According to the NCVS, black people are actually slightly more at risk of violent criminal victimization than white people (so it's not true that they're underrepresented among victims of non-lethal violent crime), but only 4% more so I think you make a good point because this seems hard to believe given how overrepresented they are among victims of homicide. I suspect that the most at-risk black people are undersampled due to racially biased non-response rates.
Robbery is extremely unusual in it's victim and offender demographics. Classify it as a crime against property, as the NIBRS now does, and you get a completely different picture of crimes against persons. In general it's a mistake to consider this issue without disaggregating by crime. Details here:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4405359
This paper looks interesting, but according to the NCVS, robbery makes up only 10% of violent crime, so I'm not sure how it could make such a large difference if we excluded it and I also don't think it excluding it from the data on violent crime is legitimate if we're trying to explain racial/ethnic patterns of victimization. I agree that it would be interesting to disaggregate by type of crime though, but I'd be more interested in what things look like once you exclude domestic violence, which is going to be overwhelmingly intraracial by its nature.
If you just look at what NIBRS now classify as crimes against persons (Assault, Homicide, Human Trafficking, Kidnapping, and Sex Offenses) I'd be interested to see the results. You probably also want to exclude simple assault, just look at aggravated assault. Robbery is now grouped as a crime against property. Grouping by motive makes a lot of sense for some purposes, for reasons discussed here:
https://www.hup.harvard.edu/catalog.php?isbn=9780674976597&content=reviews
A lot of discourse around this issue is implicitly (and sometimes explicitly) essentialist, there's no contemplation of the possibility that behavior is driven by incentives. Back in the 1890s, the Mississippi Supreme Court considered robbery and murder to be "robust crimes of the whites" so they did not strip convicted robbers and murderers of the right to vote. Instead they disenfranchised those convicted of fraud and other "furtive" offenses which they associated with Black Americans. Now the stereotypes have been reversed, but not because groups of people have magically changed character. The case is discussed in the book linked above if you're interested.
If I want to disaggregate, I'll have to use the microdata, but I don't understand the logic for excluding robbery. I understand that you made the case for grouping by motives in your book, but I don't have time to read it at the moment, so I was hoping that you could explain the logic succinctly here.
In any case, if I do take a look at the microdata to disaggregate, I'll publish some kind of update and will upload my code so anyone can easily group crimes in any way they want and check how that affects the results of my analysis.
Again, personally I'm inclined to think that people who commit robberies are similar to people who commit aggravated assault and I'm more curious to see how excluding domestic violence would affects the results, but I could be wrong.
Robbery is special beacuse it is a crime of appropriation in which victim and offender interact, so stereotypes come into play. Can't happen with burglary, larceny, or motor vehicle theft in most cases. So the victim and offender demographics are completely different. Lots of data on this based on NCVS, arrests, etc in the book, and some in the article in press. The point is this, if you want to talk about interracial crime and you really want to understand it rather then just reinforce an essentialist narrative, you have to disaggregate. Demographics of murder, kidnapping, sex crimes, aggravated assault, and arson are all very different from robbery. There's a chapter on robbery in the book, pdf available free in most academic libraries.
Rajiv -
I hope you'll see this:
Indeed. Buried in this is something that seems kinda obvious: You're going to rob someone who's more likely to have more money. That's going interact with race. I didn't actually read this post in detail so I don't know if that was taken into account but I saw your name in the comments and i know you're worth reading.
I heard you on Glenn's pod, (BTW, interesting discussion on using betting markets as a focusing mechanism in voting prediction) and found your quick mention of geographic location as a confounder on these issues. Do you have a link?
My own anecdotal experience is that property crime is vastly underreported, especially in bad crime areas. The first time somebody broke into my car in Baltimore I reported it. The cops said there was a 0% chance they would investigate and that basically they would only do anything if it was murder. I never reported any of the other break ins.
NCVS (used in this post) includes unreported crime
https://nij.ojp.gov/topics/articles/race-trust-and-police-legitimacy
But there also is most likely underreporting in some demographics due to a higher lack of trust in authority.
Shit - this belongs here:
Rajiv -
I hope you'll see this:
Indeed. Buried in this is something that seems kinda obvious. You're going to rob someone who's more likely to have more money. That's going interact with race. I didn't actually read this post in detail so I don't know if it was taken into account but I saw your name in the comments and i know you're worth reading.
I heard you on Glenn's pod, (BTW, interesting discussion on using betting markets as a focusing mechanism in voting prediction) and found your quick mention of geographic location as a confounder on these issues. Do you have a link?
More money *in their possession* and this is correlated with race but not in the way you might expect; resistance is by far the more important concern, there is good evidence for this; resistance is high when you have more money in your possession relative to total wealth; links below for more on this and the geography of lethal force:
https://www.hup.harvard.edu/books/9780674976597
https://www.annualreviews.org/doi/abs/10.1146/annurev-criminol-022422-120610
As I said before, I don't think your argument about why robberies should be excluded very convincing, but in any case it doesn't matter. After we had this conversation a few months ago, I redid my analysis using the microdata to exclude robberies and it didn't affect my argument, i. e. there are still large differences between the observed distribution of victims by offender groups and the expected distribution under the third model I consider. Obviously, the various caveats I make at the time on how to interpret this result still apply, but excluding robberies doesn't fundamentally change anything. I wanted to look at a few other things, but I didn't have time, so I didn't write a follow-up on this. But I'll try to do so eventually and I'll upload the code that allows for disaggregation by type of crime using the NCVS microdata.
Thanks Rajiv. It's interesting how Fryer is hooking up with Bari Weiss on an anti-woke campaign. I really wish he wouldn't have done that.
Fryer is not so easy to classify. His piece on DEI for example is quite nuanced and he pushed back on her blanket statements in the video. Bari herself I find to be quite reasonable on many issues and have been a guest on her podcast once. Links below.
https://rajivsethi.substack.com/p/what-is-merit
https://podcasts.apple.com/us/podcast/what-to-do-about-guns/id1570872415?i=1000565759542
I'll watch the podcast but my assessment of Bari is quite different from yours, particularly with her long-standing hypocrisy on "free speech" with respect to criticism of Isresl.
As for Fryer, while I can certainly understand his comments about karma here:
https://x.com/sfmcguire79/status/1758955868094099497?s=20
I would like to think he's rise above that kind of response and certainly his lack of response to Bari's implying that Gay endorsed "calling for the genocide of Jews" was really not even understandable.
Yes I agree about Bari on Israel, that is one issue where she won't countenance dissent, unlike Noam Dworman for example. Would love to see Norman Finkelstein or Rashid Khalidi on FP but not likely, both have been on Live from the Table.
Roland is a part time stand-up so I think he was just performing a bit but I see your point.
"The point that it was making is that, if you listen to the media, you’d think that black people are much more at risk of being victimized by white people than the other way around, but in fact the opposite is true."
I've never seen anyone in the media claim that for crime overall, blacks are much more at risk of being [violently] victimized by whites than vice versa. Is there any research illustrating that the media frequently makes this claim?
(I have seen people in the media say that blacks are more likely to be victimized by white cops, but that's not a generalization about all violent crime.)
Anecdotally, a lot of local news reporting is dedicated to crime, and it's pretty common to see Black faces when they put up mug shots of offenders. I can't speak to percentages, but the media is certainly willing to report on violent criminals who are Black.
I don't have a Twitter account, but I saw your more elaborate bar graphs for rape, robbery, aggravated assault and simple assault. Surely it can't be right that neither whites nor Asians ever rob Asians?
https://twitter.com/phl43/status/1658908170620739604?s=20
Interviews with criminals may be relevant. In Wright and Decker’s Armed Robbers in Action, they interview a few dozen (mostly black) active robbers in St. Louis (snowball sampled through a reformed ex-robber). Robbers consider several factors when deciding who to rob, and there is some disagreement among them on who is the best target. But there was general agreement that whites were safer targets — less likely to fight back. Given the demographics of local neighborhoods race probably is a decent indicator of whether someone has experience with violence or follows the code of the streets with its ethic of retaliation and toughness.