"I think there’s clear evidence that the current murder spike was caused primarily by the 2020 BLM protests...I understand this is the opposite of what everyone else says, but I think they are wrong."
You must not be reading many conservatives. Pretty much a standard gripe of theirs is that of course it was all BLM, and the media lies about it and obfuscates what's going just like they do with everything that they find ideologically inconvenient.
You make a good case, but I am left with the nagging feeling that you are affected by retrospective bias. It's not like in mid-March people said, "Pandemic! Argh! Kill! Kill!" At the time we all kind of thought we were talking about two weeks to flatten the curve. As weeks turned into months with no end in sight, lots more marginal behavior started, including the BLM riots themselves.
At least, I'd *like* to believe the BLM riots and both the overreaction and the underreaction to them were all ultimately caused by cabin fever, despair, and hopelessness, mixed with a certain amount of opportunism from some quarters.
I'm arguing from a sociohistorical point of view. I'm not going to say that evildoers get a pass because they were puppeted by the pandemic.
1. The NYC shootings graph show that shootings were steadily going up during the month of May? I see the spike you mention in the caption but it seems to have actually started in mid-May rather than May 25.
2. The Minneapolis aggravated assaults graph counter your thesis? Assaults were rapidly increasing even before the protests started; they just continued their upward trend afterwards before coming down to current levels.
I don’t consume a lot of right wing News but I do read the local newspaper. Mpls police were taking early retirement, going on disability or simply quitting in large numbers. Everyone in this area knows that the TC police are short staffed and trying to actively recruit. Those still on duty do feel unloved. It’s not just known by right wing media consumers but by everyone paying attention. Those people also know that the spike in murders is largely black on black violence.
It’s more complicated than what you are presenting here though. The scale of the riots was probably increased by the COVID lockdown. A lot of the young people that participated were available because things were locked down. The young were told they ran a less serious risk if they contracted Covid and they made up the majority of the civil unrest.
Then there was the timing of George Floyd’s death, right around Memorial Day. Weather is generally good at that time of year. If the death had occurred in February the reaction would have been less severe.
Why does an observation have to be right wing? Maybe all those mealy mouthed “it’s complicated” articles prove that the so called mainstream is in fact afraid to deal in reality. So much of reality seems to increasingly be relegated, by our self-appointed explainers, to this fantasy right-wing land. Meanwhile those supposedly in the mainstream know don’t seem to understand the depth of the groupthink they’re swimming in.
I agree that the BLM protests and follow-on effects are the most parsimonious explanation. That said:
> rising gun sales (but guns are mostly bought by white people, and so can’t explain why the homicide spike was so overwhelmingly black)
There were many articles indicating that gun sales spiked particularly for blacks. I have not done a deep dive, but just to suggest I'm not talking out of my ass:
"Although the George Floyd protests in May 2020 were the largest round of Black Lives Matters protests, there had been several previous rounds. Most notable were the shooting of Michael Brown in Ferguson, MO in August 2014, and the death of Freddie Gray in police custody in Baltimore in April 2015."
I got curious about the Baltimore/Freddie Gray numbers a while back and did a writeup.
For Baltimore there is NO DOUBT when the spike began. The cause is pretty obvious, too, though statistics can't tell us that.
There’s another possibility for the mechanism that you don’t mention. I know there’s been some academic literature on the idea that the community loses trust in police and places fewer calls to 9-11. I don’t know if this has good evidence for it or not.
"A priori there’s no reason to expect the pandemic to hit blacks much harder than every other ethnic group."
Really? There have been tons of articles about how the pandemic disproportionately affected black people. Yes, this comes from a media environment which is biased in many ways as you point out, but this claim seems quite plausible to me given that crises like this tend to have far worse effects on the economically underprivileged (as well as the possibility that there is sufficiently less trust in the medical system among African-Americans to have an effect as well).
Overall convincing article though I had matching priors previously. One nitpick:
Figure 1, weekly homicides per 100,000 city population appears to show local homicide minimums in March of 2017, 2018, 2019, and 2020, with local maximums in May of 2017, 2018, 2019, 2020. Is it possible that there is at the very least a simultaneous seasonal/weather effect? Based on that chart I can't help but think that if George Floyd died on March 1st attributed murders might have been somewhat lower, though there is likely some effect from the protests.
No other graphs in this article showed multi-year monthly murder rates so I have not referenced other data at this level of granularity yet.
I agree with almost everything in this post except for the media criticism parts. The conclusions seem very similar to this January New York Times article, for example:
My sense is that (1) most people believe the spike in murders was related to the Floyd protests, (2) most people believe that because the theory has been widely aired in the media, (3) the people and the media are almost certainly right about this.
What's much harder to say is exactly *how* the protests relate to the murder surge and what could we do about it?
1. Violent crimes rise during the summer. Following the lockdown, it's not surprising that crimes rose more.
2. I'd be interested to see deaths by economic status.
3. "but all things are smaller than other, larger things, and that doesn’t prevent them from being real or relevant." But it can mean that they are much, much less important.
4. It seems equally likely that the rises in murders are caused by the underlying factors which led to the protests. I won't insist on it, but it's at least as valid as Scott's "guesses".
This government article talks about fewer people being put in prison.
And we know drugs were still being sold. Though the death rate from drug overdose was not as publicized during the pandemic, it was higher than before. Could fights have occurred more from drugs dealing, fights over money? Overcrowding in homes from lost jobs?
Steve Sailer discusses this often. One piece of evidence I've seen him mention is the increase in traffic fatalities, which might lend some evidence to the claim that it's a consequence of less policing. [1]
I mentioned it in a comment, but you can also rule recent gun purchases out using trace data, as it takes years on average for a gun to go from purchase to crime.
Kind of feels like Scott came to his conclusion first and then is now reading the data to support that. But the data doesn’t at all look convincing to me. Clearly there was a spike around May, but the data shows it was starting in the middle of May? Floyd died on May 25, and while protests began the next day in Minneapolis, they really didn’t pick up steam across the country until a few days later. But, for example, the NYC chart shows a clear escalation that starts at the beginning of May.
Like Scott says the Minneapolis Aggravated Assault chart shows it’s obvious but all I see is that before Floyd died, cases were rising dramatically, hit their peak shortly after protests started, slightly dropped soon after and stayed at an elevated pre-Floyd rate the rest of the year.
The discussion of those NYT and Vox articles on the Ferguson Effect reminds me that so much of modern "reporting" and "fact-checking" is about determining who has the right or wrong vibes or context, rather than the facts.
There was a point brought up earlier, regarding causal factors, and it seemed quite persuasive to me as a mechanism: if the black community, due to the protests (and the various grotesque murders that served to catalyze various BLM protest waves), began to perceive the police as less trustworthy, this would increase the homicide rate because homicide is correlated to distrust in institutions. It would also explain why the increase is largely confined to the black community.
So an increase in murders after the BLM protests is an indication of reduced policing. Because there is NO possible causal path between the BLM protests (or the events that provoked them) and increased murder, except for reduced policing.
And BLM and the pandemic are the only possible causes for an increase in murder in black communities, so if it doesn't look like the pandemic, it must be BLM.
But what was causing the increase in murders? Just the reduction in police presence? I wish we had some data on the situation surrounding the murder. Was it drug gang violence, domestic violence, robbery, etc.
I suspect that attempts to claim the spike began in March in order to avoid political blame are specious, but I also have to question your implicit assumption that a spike due to covid would necessarily begin in March. What if it took a couple months for lockdowns to make everyone crazy? That could explain both the severity of the riots AND the murder spike.
Minor nitpick on the German murder graph - it looks like this is a graph of "Mord" (murder) as opposed to "Totschlag" (killing), which are separate offences in Germany. Shooting someone because you're angry with them, which is presumably what we're talking about with most of the murder spike here, would be Totschlag. Mord requires the crime to be evil in some sense (eg. treacherous, for the purpose of eating the corpse, committed with poison etc.).
Unlike most people who are reading this piece and might bring it up, I would like to say that I think the new blue color for the site is nice and feels less generic and more intentional than the previous white background.
Minor gripe: I don't think the phrase "clear evidence" should be used in cases where there's been no attempt to actually isolate a causal effect (e.g. using a convincing differences-in-differences or instrumental variables approach, or an observational approach with a real claim to causality). The methodology this article uses could provide, *at best*, somewhat weak evidence.
Major gripes: making causal claims from time series data is really, really hard, and econometricians put a lot of effort into thinking about how to do it well. This article doesn't do any of that, and the genre "rationalist wades into ongoing academic debate without avoiding any of the pitfalls the field is already aware of" worries me epistemically.
Some specifics: many of your graphs don't show what you say that show, at least not clearly enough to count as evidence. A graph which is increasing before the event isn't evidence that the event had an effect, *even if you have a story to explain why the pretrend exists*. (It isn't evidence that the event had no effect either. It just means you can't make a convincing causal claim from it.) Similarly, the story you give to the Chicago police presence graph doesn't make that graph into evidence either -- you could come up with a similar-sounding story for pretty much any random walk.
To make an "it happened in place X but not Y" claim causal, you need evidence that X and Y would have otherwise responded the same to an outside shock. This doesn't work with America and any other country in crime (so many guns! no reason to assume the pandemic would have the same effect!), and also doesn't work with the black/other races response to the pandemic (black communities were hit much, much harder!).
It's methodologically very, very hard to distinguish "the protests caused this" from "the pandemic caused a huge increase but the first few months also had a large decrease because people were stuck inside" or "the pandemic caused a huge increase but only during the summer because that's when crime usually spikes" or something similar. Doing this right would be dozens of pages of econometrics, not a handful of graphs.
tl;dr there's causal evidence that policing reduces crime, but the spikes in this article would be much larger than any of the well-identified evidence. (I'd expect maybe a 10% increase at most?) I have no idea whether or not that's because they in fact very large, or because they're the results of other causes, but I don't think this article provides strong evidence that they're the result of the protests.
I'm surprised you think there's enough evidence here to confidently state that the pandemic had so little effect. Is your view that had the BLM protests occurred without the pandemic then we would have seen a similar homicide spike? Why can't both the protests and the pandemic be important factors? Why can't the pandemic have potentiated the protests and their knock-on effects?
You cite the timing of the homicide spike vs the beginning of the pandemic many times, but that's unconvincing to me. The beginning of the pandemic was characterized by widespread lockdowns and much less activity. It was also still pretty cold in many northern cities. As you said yourself, murder rates often spike with warmer weather--couldn't the pandemic plus warmer weather potentially explain the timing? (And yes ALSO the protests?)
I also don't think it's useful to compare the US with other countries--other countries have very different homicide rates to begin with (and different social safety nets, and had different responses to the pandemic, and different gun laws, etc). To say this proves the pandemic wasn't a factor in the US specifically seems questionable.
In general I think it's harder to disentangle the various factors here than you claim.
Another data point: Miami neither had widespread riots or a major defund the police movement. (Or at least if there was one it didn't succeed.) There were peaceful protests but there was minimal rioting. Miami does have a significant Black population too. Violent crime decreased while arrests went up representing (at least according to the local politicians) the police interrupting people trying to commit crimes. This is despite not having as much of a weather change or sticking people indoors as long and having pretty widespread gun ownership and a long history of organized crime. (Seriously, it's Miami. It's still a major drug port.)
Anecdotally I lived near a lot with a lot of high end cars. The police basically had some officers watching it full time and they'd arrest a few attempted car jackers a day during the first few weeks of the protests. After a week or two the thieves (but not the protests) died down. Either due to thieves taking the hint or literally removing them from the streets. If there was a pullback I didn't notice it. If anything it seemed like the police were more active.
As one of the people who questioned your attribution in the previous article, this post causes me to take a significant update in the direction of your thesis. I'm still not totally convinced, primarily because causal inference on observational data is incredibly hard, but the evidence you present here is better than evidence I've seen presented in favor of other points of view.
I also think I was more importantly wrong on a meta point. Part of the reason I questioned your assertion is because I thought something like "this is just the sort of thumb in the eye of sanctimonious establishment media that would be catnip for Scott", so I thought it was likely that you would be biased on the subject and may be making overstated claims on questionable data. While I still think you may have some bias here, I also think your conclusion is honestly arrived at with proper consideration, and I regret any implication of unvirtuousness in my previous comment.
I predict a spike in comments elsewhere in what used to be called the blogosphere to the effect that "once again Scott Alexander has shown his true colors and good people everywhere will denounce him and never read him again."
I appreciate your analysis here, but there were numerous points that left me rather confused. First, "A priori there’s no reason to expect the pandemic to hit blacks much harder than every other ethnic group" — no? the lowest income groups in the US are blacks, Native Americans, and latinos. All of these had an uptick in the homocide rate during the pandemic (e.g. chart you present midway down). An entirely plausible alternative hypothesis is that lockdowns pushed people in low income groups into more dire economic situations, and this lead to more murders. It seems all non-white/non-asian groups had a rise in homocide rates that you're ignoring here, and class is an obvious potential confounder.
Second, a huge problem in the "Effects of other Protests" section is that protests are perfectly confounded with the underlying event perceived as a huge social injustice worth of mass protest. There are many possible causal hypotheses here, two that come to mind are:
(1) death by police → protest about policing → policy of reduced policing → more homocide/crime
(2) protest about policing ← death by police → less trust in social and government institutions, less prosocial behavior → more homocide/crime
Increase in homocides could be (1) alone, or (2)'s right path alone, or (2)'s left and right paths, or (2)'s left and right paths, with the left path also going along (1)'s entire chain to also lead to more homocide (it's hard to draw DAGs here, so hopefully this makes sense). The point is that death by police and protest are coupled, so it is nearly impossible to disentangle "the effect of protests" from "the effect of less trust in social institutions less prosocial behavior because of police murder". So I think the claim "If Black Lives Matters protests can cause homicide spikes, we would expect to see one around this time also" isn't alone sufficient, because it fails to control for the upstream cause of the protests. In reality, I bet it's both — protests reduce policing, reduced policing changes the economics of crime a bit, and less trust in our social institutions leads to more antisocial behavior. But I wouldn't be so quick to blame protests as much as what caused the protests. Protests could act primarily through an interaction in that they spread awareness of police murder and that reduces social trust much more quickly and widely than without protest.
Regardless, I do agree it is rather unfortunate the media covers these issues with such heavy partisan lenses, and not as a really tricky social problem with unknown causes that we should really work hard to understand, so we can try to have some idea of a sensible intervention.
The charts you show reinforce my distaste for rolling average plots. They often confuse when the effect actually happens. That's particularly a problem when you're trying to tease out causation, which is often what those plots are used for!
If you think your data has some artificial noise or periodicity, try leaving it in and just labeling it! I think you'll find that most people can figure that out.
I think a factor that's been overlooked is fiscal stimulus contributing to gray market demand for firearms. I've been through the legal system twice (once during and once 5 years before pandemic), and this time it def seemed like the judges were having a strong reaction specifically to the prevalence of guns at the time. Imo, everyone having more money at the same time really raises the perceived value of having a gun (either for greed or protection), and this could explain why there aren't similar spikes in countries where guns aren't readily available
I'm not following how depolicing leads to an immediate uptick in homicide. Petty crime, sure; robbery maybe; murder seems like a stretch. I agree impunity rates affect criminal behavior *on some time frame* but is the feedback that immediate? An alternative hypothesis is that the depolicing is a response to the uptick in crime caused by calls for black lawlessness combined with demoralization of policing. I would guess that the type of policing that falls the most is the kind of "circling the block in a police car type" of beat, not the difficult detective work and DA buy-in needed to bring someone in for homicide. (I know this firsthand: my brother was murdered in August 2020, and we're still fighting to get his killer arrested nearly two years later.) So I think Scott needs to go into more detail on the mechanism behind depolicing -> more murders. But I enjoyed the piece, and I agree that the NYT sucks on this.
Great analysis. I think events turned the social tenor toward chaos, and people who are only marginally civil exploited the opportunity. Why are certain races victimized, certain properties targeted? Some blame the pandemic, global warming, colonialism, police abuse and murder of criminal suspects, racism, inflation, white supremacy, etc. If one is determined to act unmoored and/or violent, to dishonor the social contract, any excuse will do. Someone usually benefits from chaos.
The response time of the homicide effect to the protest "intervention" seems awfully fast - faster than I remember the politics moving. When did police departments actually start quitting or pulling back? One could imagine a more immediate effect of "the police are dealing with the protests and not other things", if your theory is that police presence is strongly preventative of crimes. In sports you often double-cover your opponent's star player, so #2 or #3 ends up getting more plays. Maybe it's as simple as the cops having steady-state capacity but not surge capacity.
But at least for homicides the picture still doesn't quite make sense. Even though the timing looks somewhat compelling, the motive is still muddy. Why does inter-racial grievance lead to a spike in intra-racial violence? Does police presence actually work? Are cooler heads who would ordinarily prevent this violence out at the protest? Are people deliberately taking advantage of the lapse in policing to go gunning for someone? Are people of their own race just close at hand when they get back from the protest with their blood still running hot? This seems like a critical part of the story to nail down for it to emerge into that respectable conversation and out of the fringe right.
This is the big problem with the argument that "COVID was a global event, but the homicide spike only happened in America." The George Floyd/BLM protests were also a global event! There were protests on every continent including Antarctica. https://en.wikipedia.org/wiki/List_of_George_Floyd_protests_outside_the_United_States A bunch of protests outside the US got quite rowdy. Just to give one example, on June 6, 2020 the US embassy in Mexico City had to be locked down due to violent anti-police brutality protests in the vicinity that saw cars being set on fire and businesses being vandalized. https://www.newsweek.com/us-embassy-mexico-city-locked-down-due-george-floyd-protests-1509112 Yet I've seen other people cite Mexico's *lack* of a homicide increase in 2020 as evidence for the BLM protests causing the homicide increase.
And as other people have pointed out, several of your graphs show an increase starting before May 25, not after it. Here's another graph that shows the same thing, scraped from the Gun Violence Archive that supposedly covers all of the firearm homicides in the country: https://pbs.twimg.com/media/E20sKyAVEAAOAMk?format=jpg&name=4096x4096 It clearly shows an increase above 2019 levels starting in April, not at the end of May. The source is this blog post by none other than Steve Sailer, who claims it was sent in by a reader, so it clearly meets the Criterion of Embarrassment. https://www.takimag.com/article/the-racial-reckonings-new-normal-50-murders-per-day/
You might say that that still doesn't show an increase starting shortly after the pandemic. But we also have to take into account other effects of the pandemic. The most obvious is people spending a lot less time outdoors. Here's a graph of mobility in 2020 compared to 2019: https://pbs.twimg.com/media/FIxre9dVEAQKqP4?format=jpg&name=medium It shows a massive decline starting in March, bottoming out in April, then returning to 2019 levels...in the middle of May. People are obviously less likely to commit murders when they spend less time outside, so this data massively confounds any time series data on homicide.
We would expect that more consumption of alcohol would lead to more alcohol-related murders. We would expect that more consumption of illegal drugs would lead to more drug dealers shooting each other. We would expect decreases in mobility to offset this in the period immediately after the pandemic started. We would expect this to hit black people harder than other races because black people are more likely to deal drugs (and probably even more likely to be involved in urban drug gangs that are more likely to get into violent territorial disputes with each other than IE meth-dealing biker gangs). It isn't that hard to think of reasons why this phenomenon might hit homicide numbers harder in America than in other countries. So it seems extremely plausible that at least some of the increase was due to changes in the consumption patterns of intoxicating substances, and at that point why not apply Occam's Razor?
I admit that one flaw in the alcohol and drugs theory is that Mexico didn't see a homicide increase in 2020, but it's possible that COVID border crossing restrictions made drug smuggling more difficult and that offset some of the impact on the Mexican drug scene.
And here's one thing that I have yet to get a good answer for from "Ferguson Effect" advocates: why wasn't there a homicide increase in Los Angeles following the 1992 riot, which was worse than anything that any single American city has experienced since then? Here's a graph of homicides in major American cities in the 90s: https://pbs.twimg.com/media/FQxUdYVVUAImPk6?format=png&name=medium Los Angeles doesn't stand out at all.
Heritage has nothing to gain by using a less provokative title. The left have overwhelmingly demonstrated their hyperpartisanship on this issue and categorically are not interested in the truth on this issue. A strong claim, but when you have the peak "respectable" news site in the NYT basically reporting something they don't believe to be true, it shouldn't be a controversial claim.
> Murders definitely rose a little after Boudin took office, but that’s because that was also when the Black Lives Matter protests happened, which demoralized police and led to a so-far-permanent spike in murders nationwide.
That specific phrase, "demoralized police" was something that I took issue with, and made a comment on. The specific claim was not that police pulled back, it was that they were demoralized. And now in this article you say:
> I don’t want to speculate on which of these factors was most decisive, only to say that at least one of them must be true, and that police did in fact pull back.
But in the previous article you directly attributes it to demoralization, which was what I took issue with, and why I left a comment there.
The evidence does seem to support the case that the protests (and related unrest, especially in black neighborhoods) led to an increase in the murder rate. Though from the stats here I'm not convinced that reduced policing is mediating this? If a policing pullback were the cause, I'd expect a murder spike earlier march/april when policing was reduced for the pandemic response. The Freddie Gray graph also shows a more gradual decline in arrests starting well before the protests, whereas the spike in murder rate was relatively abrupt and more tightly coupled to the protests.
I think Scott makes a good case for the pullback in policing that followed the BLM protests contributing to the 2020 homicide spike. But there's another factor that may have played at least as large, and I think even larger role: the social and economic stress caused by Covid.
One good measure of this additional stress is the number of overdose deaths. This from the CDC:
"Provisional data from CDC’s National Center for Health Statistics indicate that there were an estimated 100,306 drug overdose deaths in the United States during 12-month period ending in April 2021, an increase of 28.5% from the 78,056 deaths during the same period the year before."
One can reasonably infer from this, and the increase in overall drug use it implies, that the emotional impact of Covid on the residents of low-income, inner-city areas, most of them living in relatively cramped conditions and with very few financial resources, must have been huge. The added stress falling upon a population with very easy access to guns, suffering acute financial distress and unprecedented social pressure, could have led to an increase in homicides that was unrelated to police withdrawal.
The timing of the spike in killings, a few months after the Covid lockdowns began, is at least as consistent with the evidence as the explosion in BLM street activism, which itself might have been, in part anyway, a consequence of the unparalleled intensity of social stressors in 2020-21.
Would benefit from more discussion of 2021 and 2022. Here in New Orleans the murder rate is even higher this year, which it’s hard for me to plausibly attribute to BLM protests in the summer of 2020. Also the “pandemic started months earlier” point is unconvincing, could be that it’s prolonged isolation that causes breakdowns in relations
I live in Columbus Ohio. A city not in these charts. The city population is about 900,000 and the Metro is about 2.1 million. In the city of Columbus, murders did increase dramatically in 2020 and the city set a record with 175 murders. Whether or not it tracked the timing in other areas, I do not know. Whatever the reason, the spike continued and got worse in 2021 when there were 205 murders.
Just when you thought you knew what was going on, and for equally mysterious reasons, there have been only 62 murders as of June 25, 2022, the 176th day of the year. At that rate the number of murders will drop to about 130 a 37% decrease.
There have been no changes in the political leadership of the city since the beginning of 2020.
Police responses are almost certainly one mediating factor, but let's not overlook the simplest possibility: BLM implied that police all across America murder blacks; this made a lot of people really angry; angrier people are more likely to kill people. I suspect the Marxist founders of BLM would consider a spike in homicide a feature, not a bug. It both signifies an increase in revolutionary consciousness, and destabilizes the society to be overthrown.
This isn't a tight empirical/causal argument so the level of confidence in these statements exceed the available evidence. For example, what if the pandemic caused BLM and both had contributions to homicide. The level of data storytelling presented here could be explained by this same model and we're back at square one.
It's clear that the media has an agenda - and perhaps they left the conclusion to this unusually and purposefully vague. However, I think that vagueness is the right answer to reach, however inconsistent, and "data" presented here is short of being the evidence needed to match the confidence of your claims.
I think the issue of guns is kind of too easily hand waved here.
The idea that, because most guns in the US are bought by white people they can't have an impact on a subcommunity seems specious.
For example, people buying weaponry to smuggle down into Mexico is not a large proportion of US gun sales. But, if you then said, therefore the availability of guns from the US is not a factor in making the Mexican cartel wars super violent, it would be wrong.
Similar with all those countries that aren't awash in guns that didn't see a spike in the murder rate.
I like the article and I find it mostly convincing as a whole.
One part which I don't understand is the relationship between people choosing to commit more murders and the decline in police presence/action/screening or perhaps it is the anger stirred up by the protests and angry people kill more often, just like hot people do with more summer murders. Are we just x amount of irritation away from attacking each other to death?
I get random stops by police might cause an arrest for illegal firearm possession or something which then prevents a murder by removing the gun and interrupting a crime of passion....but can that really add up to so many more murders?
I get we look at the protests and we look at the police actions...but no one is saying it is the police or protesters themselves going around murdering people. It is random other people killing more.
Do people who want to kill think they are less likely to be caught and decide to kill more often? Are we saying that police presence and actions to arrest for minor crimes are a huge deterrent and/or they interrupt such rare events by harassing the general population? Or a targeted population of young men of a certain race? It just seems like you could stop thousands of people and not interrupt or stop a single murder.
One factor which is hard to pin down is if the desire or circumstances for murder to occur have themselves gone up. Which I don't understand. The police killed a black guy one didn't know, so one is going to go out and also murder some other guy?
I don't get the logic...I saw only 3 police patrols instead of 7 police patrols in my neighbourhood over the past week....time to go kill someone?! I don't get it. Is seeing a few cops and a handful of extra guns or drugs taken off the streets here and there really the reason people choose to kill others or not? It isn't like murder isn't already very very illegal with heavy penalties. Or that a lower police patrol rate means you think the cops are going to work less hard in a murder investigation? Do killers think this?
Is a 10% or 20% less likelihood of feeling like you'll get caught the threshold for people to go out and commit the murders they already wanted to commit? Is that all that separates life from death for so many victims of murder? Are gangs just going nuts because there might be a 30% reduced rate of cops driving around?
I just feel like something is missing here. Do people want to kill more than before? Do they want to kill but feel like they were being held back and those barriers are gone? Is killing like other crime from the 'broken windows' hypothesis and simply needs a context? Killing seems like far less of a crime of opportunity than shoplifting does and context feels very weak as an explanation. But again, summer murders are higher, so maybe context matters far more than I'm thinking?
I suppose people just wanted to murder each other more in 2020. Why do americans like to murder so much more than other nations. Perhaps there are tipping points or thresholds of poverty or population or gun ownership or gang violence or drugs and despair or undemocratic abusive workplaces, but even when you 'control' using statistics for the number of guns or whatever...the US murder rates are higher. And that's not even counting the invasions, wars, or drone strikes of the american empire!
For weird reasons I've read a lot about this issue so here are a few points I want to raise wrt to your argument:
- Using only city data seems like it could be misleading if you're making conclusions about a murder spike that happened nationally. CDC data is probably more useful here if you're interested in figuring out the timing of the spike. The CDC found that murder was already up 20% nationwide in April 2020 and up almost 30% for blacks. Then there was an additional spike in June after the protests started.
- The causality of the protests on the murder spike is kinda dubious even though part of the spike does line up with the protests beginning. I brought this up in a previous post but the 2020 BLM protests saw an outpouring of young people, particularly black and in urban areas, into the streets after months of lockdown. When tons of young people start all coming out into public at once, you're going to get a rise in murders no matter whether police behavior changes or not. For example holiday weekends in the summer (i.e. the 4th of July) are often the deadliest times of the year for murder because people are socializing all over the place and you end up with tons of drive-by shootings and the like. So based on this you could explain the spike by saying that propensity to murder was already up before the protests (which I think is born out in the CDC murder stats) but lockdowns were holding people back a bit from actually carrying out murders - until, that is, the protests hit and the dam of social distancing broke.
- Interestingly murder went way up in 2020 but almost all other crimes didn't. Both police reports and victimization surveys (which don't rely on changes in police reporting) corroborate this. So this begs the question of why, if de-policing caused a murder surge, it didn't cause a surge in other crimes as well. Is murder just unusually sensitive to changes in policing, compared to other crimes? It's possible, but the difference would have to be huge to explain the lack of any changes in, for instance, theft and robbery.
Somehow this reminds me of a tweet thread I saw recently presenting a study that paid Fox News viewers to watch CNN instead. The author made it out to be about the pitfalls of partisan media in general—they just happened to pick fox because it matched the current administration. But pretty clearly the big selling point was that it confirmed liberals’ biases about right wing media without requiring them to face up to any of their own media’s blind spots.
Wonder what a study would look like that paid NYT readers to follow ACX for a while.
Has anyone bothered to see if all-causes deaths went up by the amount indicated by the increase in reported murders? Meaning, did more people die in those communities because there were more murders, or were there the same number of deaths and more were officially classified as murders?
Because, maybe I'm getting a biased view from listening to liberal-minded podcasts and interviews and articles on this subject, but it really feels to me like official crime statistics have more to do with what the police are incentivized to report than with the actual amount of crime happening. And they were certainly incentivized to report an increase in black-on-black crime immediately following the protests.
It look like the spike is something like a 50% increase immediately, dropping back to baseline over a few months. Do I think police have enough wiggle room in how they report things to invent such a spike in their statistics even with zero change in actual rates? I think it's plausible, yes - that seems like an increase of 1 homicide per 100k citizens per month, not a huge number of bodies to reclassify.
Classify some marginal things as homicide instead of manslaughter or accidental death. Does the coroner usually take 2 months to file cause of death paperwork? Get them to dump the backlog this week and speed up the process this month. Do you often ignore people who look like gang members or criminals when they try to report a crime? Take the report this time.
I feel like this wouldn't be hard to do, especially if your previous incentive was to keep the reported numbers as low as possible to look good to your superiors, so you're starting from an artificially depressed baseline.
I'm not claiming to be an expert or know for sure that this is happening or could happen, but from everything I've heard about police departments being driven by arrest metrics and community-safety metrics and how this has affected reporting over the years, I'm surprised this isn't one of the first hypotheses tested for whenever the crime numbers do something odd.
And of course the framing of this conclusion as "right wing" is a political construct in itself - assuming that suburban whites won't care why policing declined because they're so fearful and easy to manipulate (having been a suburban white surrounded by suburban whites my whole life, I'm not sure they're wrong).
David Simon, for instance, just did a whole mini-series about how, yes crime has gone up due to police protest, and how that makes police a bunch of cowards who don't deserve the public trust.
You can ideologue with any data, and I won't even be mad. The annoying bit is the strategizing. I don't care what you think, just say that thing and not the thing you think you need to say to make me think what you think.
This made me wonder about homicide rates after the LA riots. This Time Mag article points out that immediately afterwards “In May and June, only four people were killed, down from twenty-two during the same period in 1991. Drive-by shootings fell by nearly 50 percent from 1991 to 1992, and gang-related homicides by 62 percent.” The decrease is primarily attributed to the effects of the formal truce and peace pact agreement between the Crips and the Bloods the day before the riots. So this essentially supports your hypothesis as well.
In “Votes, Drugs and Violence,” Trejo & Ley make the argument that in a society with some corruption (overlap of bandits with govt&police), periods of political upheaval generate ripples through the bandit-establishment interface. This triggers turf wars among bandits, leading to increased killings. It can be as direct as “the leader of the corrupt police was promoted & left the city, so now other corrupt police and their factions in the bandit community are battling for control.) In order to have a society with fewer bandit killings, the corruption has to reach a steady state with leadership hierarchies that are clear to those involved.
Is this the mechanism by which changes in protest behavior cause increased murders? I don’t know how tightly the police-bandit connection is held in these cities.
I do wonder if, rather than A caused B, perhaps both A and B are caused by a different C, not the pandemic. Perhaps whoever organizes protests made friends among the bandits and conducted a two-pronged campaign, political street protests to weaken police plus community violence to weaken the community & establish dominance over underground economies.
The fact that you need this many graphs to demonstrate this to smart readers is the reason I'm giving up on smart people. They are worse than dumb people.
Need pages of sources to tell you that there's a correlation between a spike in the murder rate, and the most expensive riot in US history--which happened at the same time, occurred because of the police, coincided with a widely-broadcasted pullback in policing, and of which oodles of videos of the reality were available. Gunfire going off like popcorn in all directions.
Also need pages and pages of sources to tell you that women are not the same as men.
Need pages of sources to prove anything that an effete intellectual would find unpleasant if it were true.
Police do useful work *and* commit atrocities. Probably the same is true of the justice system in general. What policies would help maintain or increase the useful work while making atrocities less likely?
It seems to me like there's a great deal of evidence here that around the time of George Floyd's murder, and a couple other high-profile police killings, there was immediately after a spike in homicides.
But it doesn't seem like any of the above data proves, or even really provides strong evidence for the idea that, it was because of the specific mechanisms you state.
Here is a way to test the theory - do homicides increase after killings *of police*? I would think that actual cops being killed is a bigger cause of cops being afraid to do their job, then BLM protesters.
Deaths of police on the job went way up in 2020, but IIRC the increase was due to cops dying of COVID.
Unrelated to the content of the post, but I much preferred the plain white background to the new pale blue background. Just my two graphic design cents.
A couple of reasons for thinking covid played a sizable role in the U.S.:
When I was a kid watching Westerns, the term "masked man" was synonymous with bandit. I suspect that encouraging masked men to roam the streets contributed to the return of car-jacking during the pandemic.
Also, lots of dangerous people were released from lock-ups during the lock-downs due to fears of infections in jails and the like. Of course, this was also something the Establishment had been wanting to do since the war on mass incarceration got going about a decade ago, so it's hard to disentangle covid and wokeness.
Similarly, both covid and wokeness played a role in the war on college admissions testing: in-person testing was called off in 2020, and then many colleges announced they were getting rid of testing, to one extent or another, to fight racism. I call this the Not So Great Reset in which temporary expedients during the worse of the pandemic feed into long-term changes in directions the Woke had been wanting to go.
Portland had 4 homicides in the first 6 months of 2020 and 55 in the last 6 months. The only thing that changed from the first half of the year to the second was that BLM and Antifa became very "active" in the city so I think we can all guess why the homicide spike occurred
I understand THAT this is supposed to be some sort of right-wing point, but I don't understand WHY without some SERIOUSLY uncharitable assumptions about how the right-wingers making the point think about black people. After all, the idea of a police strike, by which police are punishing communities for their insolence by failing to properly protect them from violent crime, would seem to me to be a pretty LEFT-WING point about toxic police culture.
On a tangential note, a few weeks ago my wife called the police because there was an African man sitting in a crosswalk drinking hard liquor at about two in the afternoon. The police arrived, helped him up, and took him aside to politely ask him if he was doing OK, how much he'd had to drink, and whether he needed medical help. The man was obviously somewhat indignant, feeling racially profiled, but my wife was not the first person to call the police, because he was sitting IN A CROSSWALK and was clearly not doing great. At no point was anyone in danger of being arrested, jailed, shot, strangled, or hurt in any way, because this was in Germany, where police are generally well trained and disposed to be helpful and resolve situations as peacefully as possible. This is policing that is actually useful.
> Here’s number of arrests in Chicago. We can see that it goes way down in March when the pandemic starts and everybody (including police and criminals) are indoors, then starts going up again before the protests. Then after the protests it goes back down and stays down. My interpretation is that people complied with the strict lockdown early in the pandemic, that effect was played out by May, and then separately the protests caused a longer-term decrease in policing.
I found this unconvincing. On all other graphs the effect of the protests was immediate, but here there was seemingly a several week delay? On top of that, the downturn in arrests coincides with a downturn in the average for previous years too, suggesting that this could be a normal thing in June for some reason. (Though usually with a much smaller magnitude.)
The Minneapolis traffic stops graph also shows the downturn clearly starting before the protests, and it doesn't look like data smoothing since there's a small spike shortly before. I found it strange to see this graph presented as though it backed up your conclusion.
"I think there’s clear evidence that the current murder spike was caused primarily by the 2020 BLM protests."
I'd argue that causation went like this:
The riots start -->
The Establishment more or less encourages them as "mostly peaceful protests" -->
The police are discouraged and potential bad actors (e.g., shooters and reckless drivers) are encouraged.
Potential bad actors are encouraged in two ways: they fear the cops less so they do things like carry illegal handguns more and speed more. And they feel more exuberant because Society is telling them they are the good guys and the cops are the bad guys.
Case and Deaton coined the term "deaths of despair" for white working class losers killing themselves, dying of opioid addictions, or drinking themselves to death. The high rates of black homicide and traffic fatalities deaths after May 25, 2020 look more like "deaths of exuberance." For example, mass shootings (4 or more killed or wounded) at black social events went way up in the summer of 2020. Traffic accident deaths among blacks were 36% higher in the 7 months after George Floyd's death in 2020 than in the same period in 2019, whereas the other 87% of the population saw traffic deaths go up 9%.
Didn't a lot of cops quit as well in relation to the protests? Leading to fewer police and more experienced cops replaced by less experienced cops. I think I have read that a few times.
In 2019, Dean Baquet, chief editor of the New York Times, addressed his staff after Robert Mueller's testimony. He said that while he, too, was sad that the NYT's Plan A to get rid of Trump via RussiaGate had failed, they should not despair because the NYT was rolling out its Plan B to get rid of Trump via playing up racism.
My impression is that, for months, many Democrats saw the "mostly peaceful protests" as a way to get rid of Trump, either in November or possibly earlier in a mostly democratic color revolution. Thus, Joe Biden, who is not by nature a progressive extremist, didn't seem to recognize that the riots were bad for his campaign until the third day of the Kenosha riots in later August 2020, when he finally started calling for calm.
"The 2020 homicide spike primarily targeted blacks"
As an Englishman, it's always a surprise to hear about the depth of the systemic discrimination in America. It's not just people and institutions, you also seem to have racist graphs!
“Several people in the comments questioned my attribution, saying that they’d read news articles saying the homicide spike was because of the pandemic, or that nobody knew what was causing the spike.”
I’ve read many “news articles” which variously asserted that the spike in violent crime and the decrease in ratings for TV sports were due to the pandemic. If someone writes an alleged news article saying that, you know that he’s a transparent, pathological liar. If someone cites a transparent, pathological liar as an authority, what does that make him? An “expert”? A “tenured professor”? A close relative of mine?
Reassuring to find an independent voice on a topic that has become (needlessly) controversial. If people are being killed, when it happens is a good start in figuring out why.
I'll put forward another theory to add to the mix. Blacks in the United States are less restrained in their behavior because they feel like they are winning. Winners get to do what they like. Similarly, fans of sports teams tend to riot more after a win than after a loss.
Has anyone worked out the total number of above trend deaths for the year? Would help to know how many extra individuals died in absolute terms. Quick estimate looking at the first graph seems to be a 1.4% increase / 100 000 people x 330million citizens = ~4620 individuals, more accurate calculations welcome
It spends a whole lot of words conclusively proving that there is a strong correlation between the BLM protests and an increased homicide rate.
Then it triumphantly declares clear proof of causation.
And, as far as I can tell, there is no actual case for causation. The arrow could go the other way, both could be caused by a third or multiple other effects, or there could be circular causation.
My takeaway is that Scott is fed up with what he perceives as a liberal propensity to refuse to believe inconvenient things, and that led him to beliefs further than the evidence supports.
Explanation for the German murder-spike in 2018: The German Police adds murders to the statistic in the year they learn about them, not in the year the murder happened (since otherwise you would have to correct the statistics when it takes them a while to figure out a murder happened). In 2018 there was a trial against a man, Niels Högel, who confessed to around 100 murders he commited by killing patients while working as a nurse. This alone accounts for half the spike you see in the statistic.
While I agree with the research and the numbers, I don't think your conclusion is well formulated. Saying that "the current murder spike was caused primarily by the 2020 BLM protests" makes it sound like there is a one-step causal connection. As in the act of protesting bad police work will always cause more homicides. This kinda reads like blaming the protesters for the murders, as if protesting somehow leads to murdering someone. This can't possibly be a satisfying conclusion if your goal is to know why more homicides happen directly after BLM protests. Of course this is fantastic if all you care about is finding SOME truth to prove main stream media wrong.
You go on to mention that "My specific claim is that the protests caused police to do less policing in predominantly black areas." just to fully ignore it the conclusion, as if the police simply have no agency in this matter. What you really do here is no better or worse than NYT or Vox since all you end up doing is focusing on an irrelevant truth to strengthen your contrarian position. Anyway, thanks for linking protesters to homicides, surely nobody would use that to justify crushing legitimate important protests.
Another potentially interesting factor to consider:
Most courts closed for the pandemic. Many of them, especially in cities, didn't reopen for several years or reopened with much lower capacity. A lot fewer murder cases were being tried in court because of the pandemic. The justice system either had to detain people without speedy trials, or it had to not prosecute crimes. Neither of which are good options.
If this were the main cause of the increase in murders, I would expect a gradual increase starting in March. Instead, we see a sudden increase starting in May. So I don't think that this is the main contributing factor. But it could contribute to why murders haven't gone back down.
Maybe I missed this in the article -- is there evidence that the depolicing caused the increase in murders (rather than just both being responses to the Floyd incident)? Genuine question. In my head, I had assumed that it was something like increased anger/tension/despair in black communities as a response to Floyd's death.
If I believed the Intercept graph, the safest city in the US was Baltimore and, um, I think the rest of the data quoted contradicts that.
Also, how did Portland have *negative* murder rates? Was there a necromancer passing through who resurrected a lot of the formerly dead? There is definitely something odd going on with that graph, and I think it may have to do with "percentage change in murders" - percent change compared to when? If there were 1,000 murders in March and only 100 in April, that would give a negative percent, but it wouldn't mean that there were no murders going on.
And if I instead believe those lying liars in Oregonian media, Portland had its own little sharp increase in murders over the period 2019-2021:
"City police and officials say last year’s increase — which disproportionally impacted Portland’s Black community — was fueled by gang-related arguments, drug deals gone array and disputes among people living on the streets. In addition, the situation was exacerbated by the pandemic, economic hardships and mental health crises.
In a May statement, Portland Mayor Ted Wheeler warned residents that “groups” were traveling to the city from Washington and California to “engage in and advance gun violence.” In addition, Wheeler said “groups involved in this violence have issued an order to shoot someone in the next 30 days or be shot for not showing loyalty.”
Similar to the ’90s, Portland officials have said the city faces a rampant gang problem. Comparatively, police and residents say the boldness of shooters, their young age and the amount of shots fired surpass what they have seen before."
What I want to know is, why aren't black people vocally concerned about the rise in homicide in their communities? Could it just be that the media-favored activists don't care, but the average black person does?
Not just homicides. Motor vehicle accident deaths increased a great deal, especially amongst blacks. I plotted car accident deaths and homicides on a monthly basis from 1999-2021 as standard deviations of each groups' monthly mean to help separate seasonal effects from secular trends. It's pretty compelling. You can see both increased in 2014 and then again a great deal around May 2020.
I'd like to add one small factor: Drug trafficking is way up.
When you heard about a 40% increase in overdoses, did you consider the size of the iceberg underneath that? Anybody who ever had a habit started again during the lockdowns. With lots of demand, lots of product and lots of manpower, local dealers have been making $$$. Naturally, there has been some conflict up the chain, and anywhere there is that much $$$, there will be criminal violence.
But that is maybe 15% of the increase in crime. The police pullback is the rest. Based on conversations I have heard *IN JAIL*, amongst criminals and Sherrifs, criminals know the cops' hands are tied, they are in defensive mode. Criminals are far more aggressive from the start when the feeling on the street changes from,
"violence or shootings are certain to bring a big LEO response",
to
"Cops on duty are down 25% already, and some are at mandatory diversity seminars. 95% chance we just get away with this crime right now...." and
"...any fool has a 90% chance of getting away with murder."
This creates some real sympathy for the cops, in a soldier-ly way.
Looking at all of this data, the more interesting question to me now is related to the figure of US murder rate by year and asking what caused the murder rate to double in the decade following ~1964?
View from the inside. I'm two steps removed from several police officers, and the reduction in policing verifiably comes from these factors:
1) Cops willfully and consciously stopped going to black neighborhoods. Literally stopped responding to 911 calls there, because the risk of themselves getting roasted in the media, skewered by a social media mob, fired, and jailed was not worth the risk of going to the black area. They stated this privately, and they acted on this on the ground. A great example was the entirely justifiable Wendy's Atlanta shooting. The police officer was acquitted, but the media mob attacks on him literally caused his familial relatives to be fired from unrelated jobs.
2) Cops quit their jobs. Staffing across the country is down, especially in urban police forces that had to deal with ACAB social pressures. They decided to work lower stress jobs such as private security.
It was less of a "own the libs" thing as it was a "well if this is what they want I guess we'll give it to them" thing and a "they've made the personal risks too high to do my ordinary job" thing.
It seems to me that Americans have a lot of trust in punishment. Not all Americans, of course, but I see this weird hypnosis where the idea of punishing someone leads to believing that the punishment will have exactly the effects one wants. The unwanted behavior will go away, and there will be no blowback or side effects.
This shows up both in long prison sentences and in the left wanting to punish the police. I'm not saying punishment never works, but it's not all that reliable.
Note that more white people numerically (though not proportionally) are shot by police than black. I've been angry that police shootings of white people (some of them sketchy) aren't taken seriously, but maybe the net effect has been better for white people. On the other hand, the justice system shouldn't just be about net effect.
One more angle, the police are trained to be very afraid of the public-- as long as that's in play, it's going to be hard to get them to treat people more carefully.
> A priori there’s no reason to expect lockdowns and “cabin fever” to hit blacks much harder than every other ethnic group.
Blacks are more likely to be poor, and so live in tinier and more rundown apartments and in neighbourhoods with fewer parks, and they have more children. Sounds like a recipe for cabin fever to me. That would also explain the uptick in arrests prior to the George Floyd murder.
> No country except the United States had a large homicide spike in 2020, which suggests that the spike was unrelated to the pandemic and more associated with US-specific factors, for which the BLM protests and subsequent pullback of policing in black communities seem to me to be the most obvious suspect.
To play devil's advocate, the US arguably has the strongest culture around freedom to move and associate.
Thank you for this. Maybe you are the brave, investigative journalists I've been dreaming of? I have one question, it's a doozy and no one is bothering because the corporate media are stasi hacks. HOW MUCH DID THE "PEACEFUL" PROTESTS OF 2020 REALLY COST? One measly Axios report, estimating the damage between $1 and 2$ billion, covered only two weeks- when we all know it went on much longer. Some of these areas will never recover- and they aren't in the rich parts of town.
I don't think the police decided to relax policing in black neighborhoods as much as they were ordered to do so by people now trying to rewrite history and hide their actions.
Looking at fig. 1 - eyeballing the data, you could just as easily say that there was no spike before july 2020. The peak in may is only very slightly higher than peaks in the previous 3 years. I would have liked to see a statistical analysis of whether this peak sufficiently explained by random noice. Certainly, if you consider the last 20 years to be homogenous (probably not a perfect assumtion, but still, looking at the last 10 years seem just as arbitrary) that spike does not seem that abnormal at all.
I guess I'm arguing that the whole post may be about nothing - sorry....
It's obvious the BLM protests and riots caused the homicide spike, but it is also obvious that the pandemic caused the BLM protests and riots to happen on a scale ten times or more greater than they would have otherwise. Perhaps without the pandemic we don't get widespread BLM riots and an increase in homicide.
What does "the current murder spike was caused primarily by the 2020 BLM protests." actually mean?
If social issues had been accumulating like tinder, and May 2020 sparked the match, does that make the match the cause? If the match is the cause, then you should be able to decrease murder rates by targeting the cause... but I laugh at the idea that BLM Protests is the social issue to address in order to reduce murder rates. It is the underlying social issues that are driving these murder rates and need to be addressed. This is why I'd like to understand what it means to scott claim BLM as the primary cause. This article lacks a broader context to make the conversation meaningful or practical. The context feels like "here is my defense of my prior claim". Cool?
I see many faults with this analysis. We lack a comprehensive baseline for the seasonality of murder rates year over year. There is no explicit causation, just casual inference. I agree with the inferences but not the conclusion. And the opposing perspecties are dismissed outright.... Scott is explicitly trying to explain his perspective so this post is inherently biased rather than exploratory.
Do I think the civil events of 2020 Summer affected murder rates? Absolutely... and it happened during a pandemic at the end of Trump's presidency. Still, it feels completely irrational to try and focus on a singular primary cause to a complex social issue. Again, if "BLM Protests" were the match, how much tinder was already there and what is the real cause? What is the purpose of writing this? What is the message to be taken from it?
I was discussing this essay online and my interlocutor pointed out, and I'm copying/pasting this:
"A very simple falsification of his theory would be if the bulk of the increase was in-household homicide."
Do we have numbers for that?
(To be honest, I think that if the bulk of the increase was in-household then that opens up even uglier questions but, hey. The point of science is to go where the data inexorably leads.)
I think your interpretation of the effect of the moving average is backwards.
When using a moving average, the measurement should lag behind drivers, not go ahead of them.
Stylised example: If the murder rate is always 5/day, then on Day N the atmospheric lead concentration goes up and stays up and changes the murder rate to 10/day on all future days, then the 7-day moving average murder rate will start rising slowly on Day N+1, reach 10/day on Day N+7, then stay there.
But you use the fact it's a 7-day moving average in some of the charts to argue for the measurement starting to go up *before* the event, as far as I can see, and that's completely backwards. Moving averages can't give us advanced predictions of the things that will eventually cause them to have risen!
All round, a lot of the very good time-resolution graphs seem to show the effect starting before the protests, and even before the protested murder itself (which was May 25th, I think), more like the beginning of May. That's very confusing to me.
But it's always tough to say when a trend change actually occurs in noisy data. Oftentimes if you just looked up to Day Y, you'd say, well, yes, Day Y is high and it's been trending up since Day X, but it "trends" up and back down again all the time just due to noise, and there's nothing particularly "trendy" about the Day X to Day Y period. But the if you look onwards up to Day Z, you find out that the increase just keeps on going and going until Day Z is clearly way out of noise range and there's definitely been a start of an upward trend at some point, and then the temptation is to go all the way back to that last inflection point on Day X and call that the start of the new upward trend, even though when you'd only seen up to Day Y, Day X didn't look at all special, just one more inflection point in a noisy bounce up and down.
You discount the spike in gun sales too much. Sure the protests led to the increase in gun sales so they are correlated. Homicide rates are still high though and that is because gun sales are still high. Police do tend to pull back when convenient to set a narrative and in some countries you even see that with gang violence.
"Guns are mostly bought by white people, and so can’t explain why the homicide spike was so overwhelmingly black"
I think Scott's overall analysis is right, but this particular sentence doesn't ring true, because legally-bought guns have a tendency to make their way into the black market via theft. In 2014 Tennessee allowed residents to keep guns in their vehicles without a permit, and then in 2020 there was a huge spike in gun purchases, related to the general craziness of 2020. Here's a graph of "guns stolen from vehicles" in Memphis, TN, by year.
I'm really skeptical about one connection here. I don't think the homicide spike could be mediated by cop pullback. The timing doesn't work. Intuitively, there should be some amount of lag behind any of the reasons you proposed for cop pullback. Funding isn't cut instantly. The 'Defund the police' messaging took a little while to dominate. Police didn't instantly grow resentful; most condemned the Floyd killing anyhow, so only became resentful due to the protests. And this fits with the graph of arrests in Chicago; the arrests only started to fall off a week after the protests started.
But the homicide spike happened as soon as the protests happened. It needs to have been mediated by a much faster effect. I'd suggest the mundane 'there were just more people spending more time outside.' We know that murders in cities already go up when the weather is nice and people are outside, and go down in the winter and in rainstorms/blizzards. It makes sense that if some social effect made even more people go outside, there might be even more murders.
Eyeballing these graphs and invoking priors about crime does not make a "compelling statistical case". Correlations in timing are weak evidence for claims that one thing was "caused primarily" by another, not "clear evidence". The observations made in "right-wing sources", or at least those you repeat in this post, are also made in the other sources you link. I don't believe it's correct to say a point has "gone unnoticed" when it is taken into account in justifiably more careful causal inference.
You've shown through your use of statistics that BLM protests directly correlate to the rise in homicides in black communities in the cities where the protests took place. Great. Now how did BLM protests influence this rise? Working out this more difficult question that should produce a nuanced answer, but you'd going to need more than stats to make it clear. What you've done at this stage is give fodder to those who care to use it.
After notable incidents of police brutality, calls to 911 drop. Most interestingly, some cities have microphones designed to pickup gun shots and pinpoint them to a block. This provides useful data since the microphones will pick up all gun shots and you can look at call logs to see how often neighbors call the police when guns are fired.
The problem with the claims of the Ferguson effect is quite frankly cops do not prevent many murders. If you want to kill someone, you're not going to do it if you see a cop standing on the street, but that just means you'll do it when your target isn't near a cop. If you kill someone because you are in a fit of rage, again that often happens in places where you normally wouldn't have a cop just standing around anyway.
The Ferguson effect is not that protests cause crime spikes, it's that when trust between police and a community breaks, crime spikes because the community no longer trusts the police and will not engage them, for example, if they hear two people arguing in the street late at night. My father-in-law was Italian and grew up in Newark NJ and his motto from the streets was "you never call the cops unless someone has a knife sticking out of them"
Now I also think 2020 had another impact on crime. The pandemic hurt open air drug sales and as a result dealers tried to move in on each other's territories to maintain themselves in a shrinking market. At least from watching a lot of shooting headlines from NYC in 2020 and early 21, a lot of them seemed very much like hits rather than the peak of a generalized crime wave (i.e. someone zips up on someone in scooter, shoots him, zips away). This can be tested when there's good data on all types of crimes. Usually one would expect homicide to be on the top of a crime pyramid with lesser crimes making up larger numbers underneath. In that model if homicides increase, all other crimes should too. If, however, we find most other crime stays the same but homicides go up much more, that indicates something is going on with the distribution of crime.
Scott: When you consider murders exclusively your conclusions make sense, but when you consider the stats for other types of violent crimes, it makes less sense. If policing went down post BLM, how do explain the fact that other types of violent crime didn't rise along with the murder rates? Violent crimes other than murders have only risen slightly (too bad I can't paste graphs into my comments).
Another data point you missed is that gun ownership among blacks went up 58% in 2021, and it had been rising for several years before that.
OTOH gun ownership also went up significantly in Latino and Asian communities (albeit not as fast as in black communities). I'd be interested if we saw any increase in homicides and or crime in either of those communities — which you'd expect to see if police started patrolling less post BLM.
I don't have an alternative theory, but any explanation needs to explain why other other violent crimes haven't risen alongside murders.
"I think there’s clear evidence that the current murder spike was caused primarily by the 2020 BLM protests."
Very sound analysis, but the conclusion seems flawed. The murder spike was likely caused by the police pullback from the streets. While the police may have pulled back in response to the protests, I do not see how we could say that the protests 'caused' the police pullback (and hence the murder spike). If I call Scott a bad logician (which he is not) and he pulls a gun on me, is it right to say that I caused him to shoot me?
"I looked into Central America, and found that Honduras, Nicaragua, El Salvador, and Guatemala also all saw murders decline rather than spike in 2020."
Taking these two factors into account, it isn't hard to see why this might lead to a large increase in homicide specifically in the United States. Demand for drugs increased, but dealers who sourced their products from South of the border were unable to meet that demand. So more money went to dealers who got their drugs from other sources, and some of them used their increased resources to expand into territories of the dealers who got their product from down South. When people started to go outside again in Summer 2020, some of those new territorial disputes turned violent.
It makes perfect sense why we would see this more in the US than in other countries. Compared to other developed countries, an unusually large portion of American illegal drugs come across one particular land border, the American drug trade is unusually violent, the easy availability of guns might have something to do with it, etc. That's without even getting to the question of whether cultural or other idiosyncratic factors might have led to a greater increase in drugs and alcohol consumption in America during the pandemic than in other countries (for the record, the first Google hit for drug overdose deaths in the UK in 2020 shows a pretty small increase of just 3.8% https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/bulletins/deathsrelatedtodrugpoisoninginenglandandwales/2020 ). And it makes perfect sense why the homicide increase would be greater in demographics that are more likely to be members of drug gangs in high population density areas where we would expect territorial disputes to be more frequent.
It all seems very consistent with the available data.
But in all seriousness, I think that the refusal of so-called "expert sources" and media outlets to self-correct when they get something wrong (like their explanation for this big crime spike) is a huge problem. It seems to me like it's mostly based on ego. People in power don't have the humility to admit when they're wrong and because they hold power, they typically don't have to.
That's why you should vote for me when I eventually run as president. I promise that if any media source or person of influence makes a mistake and doesn't own up to it when new information comes out, I will punish them in a terrifying way. Egotistical narcissism is a hard personality trait to eliminate but when it is put in direct opposition to the self-preservation instinct, anything is possible.
The left doesn't look good here and they haven't for a long time. Dating back to The New Jim Crowe (2011?) I've seen a lot of slight of hand from the left, and then when someone raises an inconvenient fact, there is more smoke, mirrors, and ad hominem arguments.
The causal chain between BLM riots and the rise in violent crime seems crystal clear to anyone without an agenda. Considering the progressive effort that goes into convincing us that black white disparities are the mostly the result of a history of institutional racism (a claim that is highly disputed even among many black intellectuals), this link looks much more direct and well supported by evidence. If we can't agree on something so seemingly obvious as criminals do more crime when they know they won't be punished for it, how can we solve any meaningful problem? Isn't this a "hard conversation?"
This highlights an area that cause a lot of compassionate but logical people to question whether the left cares more about their own sacred ideas rather than the underlying problem.
"In the second graph, you can see a more traditional presentation of homicide rates, which shows them *shooting up* after Gray’s death to a level higher than they had been in the previous twenty years."
There's an alternative explanation that fits the evidence here: the killing of Floyd itself caused the crime increase by damaging trust in the police, which led to an increase in retaliatory violence.
I posted some of this as a reply but it occurred to me that it might be worth elaborating on as a main comment. The police arbitrate violent disputes. If your friend is shot and the police to deal with it, there's probably no further violence: the perpetrator is arrested and sent to prison, end of story. But if you don't trust the police to deal with it - because there has been an extremely prominent example of them being untrustworthy - you might decide to take matters into your own hands and seek violent revenge.
This is just as compatible with the timing of the evidence you've described, including the pre-pandemic stuff in Baltimore and Ferguson.
This argument could be strengthened or weakened by research on how trust in the police changed among young black men, especially those adjacent to or involved in gangs.
I did see another comment mentioning that reported trust of the police among white people was affected more but at the risk of speculating a bit, even if a white American reported trusting the police less post-Floyd, I'm not sure they would turn to revenge violence in the event that they were the victim of a crime. Given that retaliatory violence is already common in gang disputes, I think it makes sense that it would go up there and not suddenly appear in other parts of society.
This old Tabarrok papers cleanly explains a mechanism through which we should expect riots to lead to permanently higher crime rates. Having read it long time ago, when I saw these riots I was pretty confident that they would lead to permanent increase in crime. As a bonus its explanation is less morally charged than your proposed mechanisms.
The thing is, given constant police resources the equilibrium crime rate is not self correcting. Exogenous temporary shocks to total crime will lead to a permanent change in total crime.
This happens because the decision to commit crime depends on the share of crime that is caught, but with constant police resources the share of crime that is caught depends on total crime.
So what happens is a temporary exogenous rise in total crime leads to share of crime caught falling, this makes people more prone to commit crimes, and that makes crime rise even more. When the exogenous shock is gone, there are still more crimes than before, so the share of crime caught is still lower, perpetuating the new higher crime equilibrium.
(Incidentally that is a minor plot point in True Dectective's second season, where a group of mostly law enforcement officers take advantage of the LA riots, and the consequent straining of police resources to make a hit on a jewelry store)
An interesting question would be how the homicide rate moved in cities that previously experienced the Ferguson effect compared to large cities in general.
Baltimore, for example, saw a spike in homicides after the Freddy Gray riots in 2015, which has remained until now. Notably, unlike the rest of the country, it did not have a homicide spike in 2020. Total homicides in Baltimore actually dropped in 2020.
This fits nicely with the thesis of this article. Baltimore has experienced the effects of the effects of the hypothesized depolicing, the George Floyd riots thus had little additional impact.
Edits: 1 ) removed grammar mistake, 2) clarified in the first sentence that I'm objecting mostly vehemently to Scott's characterization of the criminological literature regarding the OG Ferguson effect (see his section titled "Effects Of Previous Protests") which serves as kind of a robustness check to his claim that protests -> homicides.
Scott, you are wrong (about the original Ferguson effect, which serves as a robustness check of your more general claim of protests -> homicides)! And I can show you why. Why should you read this random comment among a sea of meshugas? Argument from authority: I'm a PhD candidate in one of the top criminology programs. In other words, I'm intimately familiar with both the content area and the issues of time-series analyses. I'm happy to privately verify this and/or share full text articles I source below if they are paywalled. Key points:
0) Begging the question and also not citing relevant literature that tests+disproves your claims
Throughout this letter (which should be revised to become a mea culpa), you conflate two separate hypotheses:
a) There is a POSITIVE ferguson effect (aka protest effect) on DEPOLICING (aka less 'active' or 'proactive' policing)
b) There is a POSITIVE ferguson/protest effect on violent crime, PARTICULARLY HOMICIDE
Hypothesis A appears true: police do pull back from policing when facing public scrutiny. The extent of this pullback is debated. For more info, see MacDonald 2019; Marier and Fridell 2020; Mourtgos, Adams, and Nix 2022; Rosenfeld and Wallman 2019. However, you beg the question by showing hypothesis A is true, and claiming hypothesis B is also true!
Hypothesis B is plausible and worth exploring! This is because at the extremes both 1) more and/or well targeted police(ing) reduces crime (Evans and Owens 2007; Sherman 2022) and 2) poorly targeted and/or no police(ing) increases crime (Loeffler and Braga 2022; Piza and Conneally 2022). Deterrence - particularly versus nothing - does exist (Nagin 2013)!
However, the problem is that we have actually tested hypothesis B and found it lacking! See the Criminology and Public Policy Volume 18 Issue 1 (2019), particularly Rosenfeld and Wallman 2019. The issue is that there's an assumption that arrests (and other CJS interventions) have a UNIFORM marginal effect, but this may not be the case. Instead, imagine that the marginal effects of incrementing/decrementing a(n) arrest/imprisonment/search/stop etc. is conditional on the cumulative exposure, much like a laffer curve! In this case, it is not so clear that reducing police interventions would result in increased crime, as long as the reduction isn't 'too large'. See Owens 2019 for a very good short summary on this economic perspective on CJS interventions.
So, TL;DR hypothesis A doesn't mean hypothesis B, and in fact we (criminologists) have already done this work and found hypothesis B lacking.
1) Causal and validity problems - aka time series analysis is hard
You've done some work, but it isn't even close to enough. To help get there, read the very approachable McCleary, McDowall, and Bartos 2017, with particular attention to the chapter on construct validity.
2) too high an expectation of there to be a parsimonious explanation - aka homicide is a highly contingent social event
It shouldn't be surprising that a reasonable desire for a parsimonious explanation - like protests -> homicide - falls apart. Why? Because homicide is actually the culmination of highly complex social phenomena. You have to have 2+ people, a place, a social context, a particular reason, and (usually in the US) a gun! Parsimonious explanations seldom provide enough of those elements.
3) Why does it matter if you are wrong?
Well, because being wrong about attributing the rise in homicide constitutes perpetuates a historic trend of perversely individualizes general social problems, creating variations of the following viewpoint (not to mention Sailer in the comments of course): 'The reason black people suffer in the US is that the poor and/or black community is composed of individuals who have low self-control, make poor decisions, have broken families, bad cultures, excessive disobedience, immoral norms, and self-perpetuating poverty. Therefore poor and/or black people profligately inflict crime (particularly bad ones) both within their community and across the US.'
Basically, this view of the world is like watching a cut of The Wire that only features drug busts and scenes of violence (aka the media up until 5-10 years ago). This means that structural conditions are ignored, or perversely endorsed in a 'race realism' mode to morally justify inequity. This is a lazy interpretation and it's uninspiring you stan for it when, in other arenas, you seek novel explanations. Like, instead of this hogwash you could write about when regulations and higher education intersect with negative externalities (e.g. Lovenheim and Owens 2014).
Bibliography:
Evans, William N., and Emily G. Owens. 2007. “COPS and Crime.” Journal of Public Economics 91(1):181–201. doi: 10.1016/j.jpubeco.2006.05.014.
Loeffler, Charles E., and Anthony A. Braga. 2022. “Estimating the Effects of Shrinking the Criminal Justice System on Criminal Recidivism.” Criminology & Public Policy (online first). doi: 10.1111/1745-9133.12588.
MacDonald, John M. 2019. “De-Policing as a Consequence of the so-Called ‘Ferguson Effect.’” Criminology & Public Policy 18(1):47–49. doi: 10.1111/1745-9133.12430.
Marier, Christopher J., and Lorie A. Fridell. 2020. “Demonstrations, Demoralization, and de-Policing.” Criminology & Public Policy 19(3):693–719. doi: 10.1111/1745-9133.12492.
McCleary, Richard, David McDowall, and Bradley Bartos. 2017. Design and Analysis of Time Series Experiments. New York, NY: Oxford University Press.
Mourtgos, Scott M., Ian T. Adams, and Justin Nix. 2022. “Elevated Police Turnover Following the Summer of George Floyd Protests: A Synthetic Control Study.” Criminology & Public Policy 21(1):9–33. doi: 10.1111/1745-9133.12556.
Nagin, Daniel S. 2013. “Deterrence: A Review of the Evidence by a Criminologist for Economists.” Annual Review of Economics 5(1):83–105. doi: 10.1146/annurev-economics-072412-131310.
Ouss, Aurélie. 2020. “Misaligned Incentives and the Scale of Incarceration in the United States.” Journal of Public Economics 191:104285. doi: 10.1016/j.jpubeco.2020.104285.
Piza, Eric L., and Nathan T. Connealy. 2022. “The Effect of the Seattle Police-Free CHOP Zone on Crime: A Microsynthetic Control Evaluation.” Criminology & Public Policy 21(1):35–58. doi: 10.1111/1745-9133.12570.
Rosenfeld, Richard, and Joel Wallman. 2019. “Did De-Policing Cause the Increase in Homicide Rates?” Criminology & Public Policy 18(1):51–75. doi: 10.1111/1745-9133.12414.
Sherman, Lawrence W. 2022. “Goldilocks and the Three ‘Ts’: Targeting, Testing, and Tracking for ‘Just Right’ Democratic Policing.” Criminology & Public Policy 21(1):175–96. doi: 10.1111/1745-9133.12578.
Gell-Mann Amnesia runs both ways. I've found claims about what "The Media" says to nearly always be false or cherrypicked when I bother to check them. (Also, is The Economist not part of "The Media"? If it is, then citing their graphs here is a self-defeating argument.)
To throw in another theory for the general situation, maybe double time for overtime is a bad incentive. Perhaps the police are too sleep-deprived to be thinking well. I've never seen discussion of when dubious behavior happens compared to number of hours worked.
People wanting to party after bars closed, restaurants. No fun. A riot was a big party where anything goes. For that crowd, police officers would be in danger. Gangs could kill them. All the prisoners being released would likely go find those who got them incarcerated. Deadly game for officers, families, informants. All people who needed to make a living suddenly hitting the streets. Many may have died from drug overdoses.
I'm inclined to think the decrease in policing was a large factor---and indeed there's a large literature showing that increasing policing decreases crime---but it does seem like there are some bugs in the argument here.
It seems implausible that BLM protests would cause a rise in people to get into fights on airplanes. Given that we know anti-social behaviour generally went way up, the simplest explanation for the murder rise is that it's the right-tail of the general phenomena. The concentration of rise in murders in black communities could be explained by a model that just says everyone now commits crimes 1 degree more severe than they were pre-covid.
That leaves the questions of why the murder phenomena is America specific, and why the violence spike started in May. One possible answer: CARES act stimulus checks were sent out in April. This large cash infusion may have enabled many relatively poor people to buy (additional) guns.
My point here is not that this is a better theory for the murder spike, but just that it fits with the main points of your argument and is additionally consistent with the general rise in (non-murder) anti-social behavior.
They let offenders out of jail and prison that normally would have been in cages. Most offenders are black. They went home and killed people. Very simple.
It was unemployment. Peaks of unemployment correlate with peaks of homicide. There are other long term trends that raise or dampen the height of the peak, but unemployment seems to me an obvious explanation.
Here's my rough sketch of the causal mechanism:
Some say 5-10% of young men (mainly) are strong, impulsive, angry and tough and not very skilled. Many times they are marginally employed. You get covid and this group gets thrown out of work en masse.
What do they do? They get into all kinds of trouble. They get into fights with one another. Some of these lead to homicides. Mystery solved.
Here's the unemployment chart since 1960. Notice how the peaks overlap with the murder chart for the years prior to 1992?
Then after 1992, when the great unleaded gas effect washed over, the peaks in homicides became *much, much* smaller, but still correlated with unemployment.
However the peak in unemployment in 2020 was off the charts, and so was the blip up in homicides.
Increases in police retirements and difficulties in recruiting tends to support the points made regarding police reactions to Ferguson, Grey and Floyd effects. What is worth recking you retirement, your life, your assets or even criminal prosecution? At some point you have to be practical.
First, what is this "elite establishment" that you refer to? The WSJ, the NYT, or USA Today, or all three or are you referring to news networks like Fox or Yahoo News?
Your calling a "protest" a "riot" already highlights a bias. Protests can become violent, but riots start out that way--I should know, I worked in a prison and went through several riots and I can testify that I would never call them protests, but you seem to have no problem calling a protest against police violence on black people a riot. Is it because that's all black people can do? And was it "your city" being looted? That sounds like something a white person would say while people of color facepalm in the background--I saw the same thing with Occupy where suburban white kinds would march in some inner city neighborhood they didn't know jack about while shouting "Whose streets? Our streets!"
Sorry, I'm not intentionally trying to be confrontational, but I thought that this site was more about using rational means to address problems.
The argument here is that BLM protests caused changes in police behaviour which in turn led to the homicide spike, while covid and restrictions did not in turn lead to the same.
BLM caused the spike, covid not. Timelines don't match up otherwise. So far pretty convincing!
At the same time, though, we can see that covid restrictions did lead to less policing, just like BLM, without being correlated with any homicide spike. Wouldn't this imply that BLM caused the spike by some other mechanism, or at least that there is some meaningful additional variable making a difference in how police withdrawal affected homicide rates between the spring and summer?
It's not totally off tye wall to use the weather to explain the increased severity of the protests/riots. The February Revolution happened because St Petersburg had a "warm" spell in the middle of a general workers' strike, and at the same time as International Women's Day. If the weather had stayed cold in February 1917, the protests would likely have been less severe and the Tsar may have held on a few months longer. If Floyd had died in February or March 2020, maybe the protests are less intense.
But the weather itself can't explain the sustained levels of violence. 2020 didn't see a spike on Memorial Day followed by a decline to pre-2020 levels. Murder rates went up and stayed there. And murder rates have retained their seasonal pattern (higher in summer than winter), which is almost-but-not-quite counter-cyclical to covid and covid restrictions. If the spike in violence was simply an outburst of suppressed energy from lockdowns, then lockdowns would decrease the murder rates and ending the lockdowns would cause a temporary uptick. That isn't really seen in any of these cities.
On the comparison to other countries - I can attest to the strictness and insanity of Covid rules in Germany, as I spent the entirety of the pandemic there with occasional trips back to the USA. Lockdowns in Germany were far stricter and longer lasting, and letting up lockdowns didn't lead to mass numbers of shootings or stabbings. It did lead to a lot of outdoor celebrations with masked people dancing 2 meters apart from each other, but that was about it.
"Police interpreted the protests as a demand for less policing, and complied."
Coming from a family of New York City policemen, this reason caused great laughter around the table. The police do not comply with the demands of protesters but The second choice also seems wrong
"Police felt angry and disrespected after the protests, and decided to police less in order to show everybody how much they needed them."
Yes to 'angry and disrespected', but the decision by the police was to punish, to avenge, and in doing so to violate their oaths.
I did a search on the murder spike in this period in the NYT:
“ Changes in policing. The fallout from the 2020 racial justice protests and riots could have contributed to the murder spike. Police officers, scared of being caught in the next viral video, may have pulled back on proactive anti-violence practices. More of the public lost confidence in the police, possibly reducing the kind of cooperation needed to prevent murders. In extreme circumstances, the lack of confidence in the police could have led some people to take the law into their own hands — in acts of street or vigilante violence.
The timing supports this theory, with homicides rising unusually quickly shortly after George Floyd’s murder and the ensuing protests. Killings also spiked in 2015 and 2016, after protests over policing during those years.”
“ The protests that erupted after the police killing of George Floyd in Minneapolis were also an important factor, in 2020, although experts differ about why. Some argue that the police, under intense scrutiny and demoralized, pulled back from some aspects of crime prevention. Others put the emphasis on the public, suggesting that diminished respect for the police prompted more people to try to take the law into their own hands.
“The distrust of police, the low morale among police, the fact that the police are being less proactive because they are legitimately worried about being backed up by their superiors” were all contributing factors, Mr. Winograd said.”
Something I've always wondered about is why there isn't a massive push to hire a ton of black cops. That should help a lot right? After BLM there are extremely aggressive pushes for affirmative action in academia, the media, and so on but not in the one place where it's really needed. Electing Eric Adams mayor of NYC was a good start. But society actually has a compelling interest in the demographics of the cops matching the demographics of the population in a way that it really doesn't with the demographics of say, scientists. I'm generally not a fan of affirmative action but I would probably support black quotas in police departments at this point.
Scott, you said this "My specific claim is that the protests caused police to do less policing in predominantly black areas. This could be because of any of:
...
Police worried they would be punished so severely for any fatal mistake that they made during policing that they were less willing to take the risk" <-- I would add a related option... Police worried they would be punished so severely for any action, even if legal, they were less willing to take the risk.
Even shooting someone grabbing your gun gets your picture in the paper, a long suspension, protestors outside your house, and the mayor condemning the white supremacy of your profession. Who wants to stop a minority and frisk him, even if you can see what looks like a gun? People are going to start filming and chanting as you do it.
“ This study investigated whether homicides increased after protested police-involved deaths, focusing on the period after Michael Brown’s death in Ferguson in August 2014. It also tests for effects of legal cynicism by comparing effects in homicide and aggravated assault on the assumption that reporting of the latter is discretionary and police abuses may make communities reluctant to notify police. Using FBI data from 44 U.S. cities, homicide and assault rates from 2011 to 2019 were analyzed using an interrupted time series design and combined in a meta-analysis to calculate pooled effects. A meta-regression tested effect moderators including external investigations and city/county sociodemographic characteristics. With a conservative threshold of p ≤ .01, 21 of the 44 cities experienced a significant increase and one had a significant decrease. The pooled effect was a 26.1% increase in the homicide (99% CI: 15.3% to 36.8%). Aggravated assaults increased above baseline, though the effect was 15.2 percentage points smaller (99% CI: –26.7 to –3.6) than the effect in homicides. When outcomes were measured as percent change, there were no significant effect moderators, but when measured as absolute change, homicides increased to a greater extent when the death was subject to external investigation and in cities with higher Black populations, poverty rates, and baseline homicide rates. The findings suggest that protested police-involved deaths led to an increase in homicides and other violence due to the distrust fomented within the very communities whom police are meant to protect.”
If policy Q leads to significantly more black people murdered compared to policy P, then my respect for the substance rather than the rhetoric of anti-racism seems to leave "more P" as the only morally viable choice to support.
Can both protests and homicide spike amoung blacks be caused by the same event? Obvious hypothesis is that the outrageous murder made lots of people very angry, some people ventilated this anger through protests others in a different way, including other murders. The distinction here is pretty important because in one case we would like to have as less protests as possible even when outrageous things happen, while in the other we would have the opposite incentive.
I don't think there's any better way to convince or persuade (those that can be) than to calmly, and repeatedly, 'look at the facts', "insist on thinking on a completely literal level", and walk thru careful reasoning about what's most likely true.
Have not been here in some time. Began reading the post and immediately felt a cringe. Seems all too common these days is to present various data to support various conclusions that are left hanging without context. And thus, we've generally become mistrusting of any presented data, assuming (often too quickly) a cynical attitude about where the presenter intends to lead us. Maybe that’s the point?
I'm not sure what conclusions we are intended to draw from this presentation, but I fear that any conclusion drawn has the potential to reinforce very negative stereotypes... in this case regarding race. Can this post lead to a more insightful and nuanced discussion about why murder rates rose in these areas, in these communities, that includes a comprehensive and objective accounting of this nation's history toward black communities? Might also be worthy of discussion to understand how a generalized fear pushes white folks to embrace these stereotypes so readily.
Why bring in the RK riots --which actually were riots!--when the BLM protests had a grievance and an organization behind them?
RK riots were confined to one location. The BLM movement had 7305 events in various cities across the country, the vast majority of which were peaceful.
Do the Right Thing was a movie, not RL. Spike Lee's character thinks hard for a moment before launching that trash can through Danny Aiello's store front window. We could debate what Lee is asking his audience to consider about the urban black's situation and race relations during this period when he performs this act, but that's as far as I'd go. You seem to be suggesting that art is imitating life and so we can use this film to what? Draw an analogy not to the motives behind BLM, but to justify using the word "riot"?
I'd argue that Spike Lee's character is provoking a riot out of frustration. This was the 80s. Today, there's enough statistical evidence and terms (i.e., structural racism) to create an agenda, demands, and a call to action.
Employing a word like "riot" when referring to a complex situation that has a history of grievances attached to it seems to me to be a way of dismissing that history and those grievances--we were listening until they rioted, now we know that we don't need to listen. Riot puts all of the weight on the black people attempting to address an existential problem that they now have the tools to articulate, but when you employ this word you are relegating them to all action and no thought.
Were there pockets of violence during these protests? Yeah, sometimes, but the larger movement eschewed that violence.
I'm not clear on how you are exercising "linguistic valence" here when you are using the very word that conservative news organizations picked to describe the BLM protests.
Regarding "linguistic valence," I'd say I'm operating in that field by addressing the RK riots as riots while seeing the BLM events as something more complex and thoughtful. You just keep using the same word.
Hmmm, there must be more to this but rather amazingly the effectiveness of police to solve murders has been droping. It took a hit in 2020 but more amazing to me is that it's been declining since the 1960's and 70's. In particular, we saw police buildup in the 90's while at the same time tools to solve murders have gotten better and more available. A lot of stuff we take for granted today like cameras almost everywhere, people's cell phones producing logs of their geographical locations over time, DNA testing, a lot of forensics, this all should have made it a lot easier to solve murders. Yet police effectiveness here seems to be falling. Are murderers getting more cunning, super-villian types that only a Bruce Wayne or Sherlock Holmes can figure out? I doubt it.
If police effectiveness is going down, then any theory that BLM protests increases crime by causing police to pull back is flawed. If police are less effective, then it matters less and less that they pull back.
I suspect what you're seeing is a gradual decline in social trust, in particular between communities with high crime and police. The reality is probably half of murders are brain dead easy to solve, the police literally roll up on the killer covered in blood holding the murder weapon. Beyond that social trust does a huge amount of work AND things like CSI science can only substitute for that at the margins. Incidents like Floyd and the pandemic accelerated and confirmed that trust was lacking.
Less that BLM protests magically cause crime, more like they are both symptoms of the same underlying dynamic.
Whether decreased policing leads to increased crime, and if so the effect size and exact type of policing that reduces crime, is a pretty open empirical question. That said, regardless of the causal mechanism, it sure seems like the crime increase of the BLM protests outweighs the minor police reform that followed (though perhaps it sets the stage for more police reform down the road... long-term consequences are always intractable). This is not to mention the political consequences of Democrats losing votes due to Defund and thus an increased chance of our democracy collapsing, less climate change action, etc.
This was not obvious at the time though. There was a brief period where it looked like we might be getting actual police reform: Minneapolis looked like it was going to remake their police force, a bunch of red states passed (very) minor police reform legislation, etc. Support for BLM and racial justice went significantly up for a brief period, figures which had been stable for a long time.
What does this means for future BLM protests? It is clear that if you are going to have them at all, make them super policy-focused, to actually get something for our troubles. But at this point it's probably not worth having them at all in the foreseeable future, as all they do is polarize the debate around left-wing Defund.
It was pretty clearly the pandemic lockdowns. It makes sense that crime didn't spike immediately but instead took a couple months before people really got cabin fever.
AFAICR this was the first time that a jury decided that a Mpls police killing was murder.
A couple years later, Derek Chauvin was the second.
Really, the thrust of this post seem to imply that the BLM movement is some sort of first cause. There was a long history of fear and anger about unequal treatment of non whites by Mpls police.
It became common in the Twin Cities to hear of a police shooting and respond with the question “Driving while black?”
Again I’ll say the violence that followed Floyd’s death was not justified. A lot of the people taking part didn’t give a damn about Floyd. There were many people - mostly white - taking advantage of the opportunity to indulge any malicious impulse that occurred to them. The police were in a crouch, all bets were off. What better time to plunder the OxyContin shelf at the local pharmacy? To pin all of that behavior on the BLM movement is just wrong. One of the felony conviction from the early period was against a masked white biker knucklehead smashing storefront windows with a 4 pound hammer, just out for a little recreational mayhem. Another was a white kid who was so dim that he brought police tactical gear that he looted from a police station to work to show off as souvenirs.
Hell, the only houses in my rather sedate St Paul neighborhood without a “Black Lives Matter” sign on their lawn are mine and the black couple across the street.
Murder increased because the police pulled back after the George Floyd riot? Yes, I already knew that and fully expected it.
Cost-benefit calculations should not look at only the costs and not the benefits. If your proposed mechanism of action is that the protest caused reduced policing which created a cost in the form of a spike in murders, then I think it would've been cooler if your analysis also included the benefits of reduced policing. What are the costs of policing?
For instance, if you're using reduced arrests as your proxy variable, we should ask how many arrests are we willing to pay in order to prevent one murder? In 2019 USA there were 10,085,207 arrests and 16,669 murders. In 2020 USA there 7,632,573 arrests and 21,570 murders. The ratio of the differences come out to 500 arrests per murder.
Is that worth it? Maybe! Depends on how bad arrests are compared to murders, and what other metrics are at play. I'm not claiming to know the answer, I'm just saying that I wish that this analysis had touched on relevant questions like this. (I get that you're responding in the context of countering the media and all that, but still ... I really feel that considering costs and benefits together and not separately is a good principle to stick to.)
This is just a back of the envelope calculation of course - a real analysis might look at stuff such as prison-life-years welfare metrics, economic losses, etc. And of course, none of this is really establishing causality super well. I'm just making a more general point that even if we grant your model of this mechanism, in order to draw useful conclusions from this analysis about whether the protests are good or bad, the costs of the intervention (here, the protests) must be considered in conjunction with the benefits.
I think it's really, *really* irresponsible to express "homicides spiked after the BLM protests" as "BLM protests *caused* the homicide spike." For one thing, there is no way to demonstrate that homicidal-ness somehow increased due to BLM (I seriously doubt that it did), which is what that assertion implies. Moreover, you say yourself that a police pullback led to the increase. So what gives? You already identified the real answer: police stopped doing their jobs. How is that BLM's fault? Cops were never seriously threatened during the protests. As someone who was at them, it was very much the other way around. You've laid out a story where cops were fleeing well-deserved scrutiny, and the result was less deterrence of crime. But this isn't BLM's fault, this is the cops' fault for being bad at doing their jobs under scrutiny. Phrasing the headline as if it was the fault of innocent protesters that crimes *unrelated to the protests* went up is incredibly irresponsible and a massive show of disrespect to people like me who want to see the policing problems raised by BLM taken seriously.
I have a friend who claims that one of the reasons that the BLM thing blew up so much, at least in the sense that left-leaning institutions like academia and the media talked about it, is that the Democrats wanted to use it to increase black turnout and oust Trump. How plausible is this hypothesis and how would we test it?
Calling the BLM protests "the cause" feels like a bit of a stretch - yes, the data shows good support for "contributed". But either one plays the cause game all the way back (police brutality leads to protests etc.), or one acknowledge that the police making agentic choices is closer in the causal chain to the measured outcome.
When you have a causal chain, what about the BLM protests specifically makes it the thing to single out in the chain? My guess is that there is some empathy and less agency attributed with the stance "of course the cops reduced their policing"; but "of course they protested" has less support. Why?
Seems like police behavior in response to the protests as a plausible theory of the cause in the spike of inter-personal violence should be addressed in a lot more detail within this essay that concludes it probably was the protests themselves. That, after all, is what the Ferguson effect is supposed to refer to: People protest the police, and in response, the police refuse to do their jobs. This reduces the deterrent effect of proactive policing, which leads to more crime.
The lesson of such a theory isn't necessarily to stop protesting police misconduct or problems with criminal justice. You could instead try to break the chain of police engaging in traditional striking behavior in response.
Of all the foreign countries mentioned Honduras is the most instructive. Honduras has the highest highest documented murder rate in the world. Other places are probably more violent but they are so chaotic reliable data are hard to come by. In response Honduras banned the private ownership of any firearms. The murder rate continued to climb. This debunked 2 libtard fantasies. That gun control works & that government can make something disappear by banning it. The government tried to ban alcohol. That worked out great didn’t it. The government banned heroin well over a century ago. You can’t find that stuff anywhere can you. Ok it’s been largely replaced by fentanyl an even more dangerous drug but the point remains.
I am wondering about the use of the word "caused" (rather than, for example, "triggered") by the protests. The post unpacks a potential complex of interrelated phenomenon under the heading of the Ferguson Effect that would explain how anti-police protests set off a set of actions that result in less policing (and possibly less cooperation between communities and police) and other effects. So the pattern would be expected. But to say "caused" carries the sense of a more direct outcome rather than a distal outcome. The difference is significant if one is thinking about how to intervene in a social pattern or whether intervention is possible.
One of the many difficulties with crime data and making inferences on it is the chain of causal effects that occurs before data is collected. For any given effect seen in the data, you generally have to ask: is X going up because more people are doing X, or is X going up because police are paying more attention to X? Murder is actually just about the only crime we can undeniably say is happening more or happening less at any given time, simply because it comes with a body and we're generally pretty good at reporting on bodies. We can say, undeniably, there was a spike in murders after the George Floyd protests - you can literally just look at the data and know for a fact that yes, that happened.
It's everything *else* that you can be much less sure of. Sure, the spike happened *after* the George Floyd protests... but was it *because* of them? That is a much, much harder question and I don't think you've at all answered it here.
There's a few questionable assumptions and assertions being made about the data here. But ignoring all of those and taking absolutely everything said here to be true, are you sure you've drawn an exclusive causal chain? Because I think I could lift most of this evidence as written and instead make the opposing claim: the stress of seeing a black man brutally murdered, in public, by an authority figure drove an increase in homicide nationwide, predominantly centered in black communities where that stress was felt the most. We need not invoke protests or a hypothetical 'police pullback' at all.
I guess what I'm saying is: did you find the data because you had a viewpoint you wanted to prove, or did you let the data lead you to the viewpoint you're expressing here?
The fact that empirical researchers are hesitant to draw a causal conclusion on what caused the spike is specifically because that's the *right* way to interpret data that is literally inundated with idiosyncracies during a year that was itself literally inundated with novel events.
I don't understand why you're so focused on policing changes as causal. If you're right that the increase is predominantly black-on-black in significantly black areas, it seems like level 1 analysis should start with.. black people, not cops. And if black people were more pissed off and angry- not because of the protests, (largely) not because of BLM, but BECAUSE OF WHAT INCITED THE PROTESTS TO BEGIN WITH- then you'd expect more violence ceteris paribus.
"Cities where large numbers of protesters turned out, relative to their populations, decreased police expenditures by an average of $12 per city resident. Los Angeles’s relatively high protest rate translated into a $2.50 per-capita cut in police spending. Minneapolis had an even higher protest rate of 77 participants per 1,000 residents, which coincided with a $32.50 per-capita chop to police spending. (For reference, cities spent an average of $422 per resident on law enforcement in 2021.)"
People keep saying things like “oh, I liked when he made an compelling statistical case showing that the media was completely wrong on this one thing they sounded very confident in, but then he started saying the media is often wrong and biased, and that sounded cliched and conspiracy-ish and right-wing, so I lost interest”.
People keep saying a lot of things that are wrong, so don't worry about it. If you sound conspiracy-ish and are correct, so be it. In other words, I think it's best to optimize for accuracy rather than for telling people things that won't put them off. We get too much of the latter already and not enough of the former.
What this doesn't really address is the surge in violent crime in rural areas:
-----
Homicide rates in rural America rose 25% in 2020, according to the Centers for Disease Control and Prevention. It was the largest rural increase since the agency began tracking such data in 1999. The CDC considers counties rural if they are located outside metropolitan areas defined by the federal government.
The rise came close to the 30% spike in homicide rates in metropolitan areas in 2020.
The CDC hasn’t analyzed 2021 homicide data yet. In some rural counties, murder rates remained high last year, while in others they have begun to recede along with Covid, data from local law-enforcement agencies shows.
...
In cities, law enforcement and civic leaders have blamed the increase in violent crime on factors such as police pulling back after racial-justice protests, the proliferation of guns, initiatives to release more criminal suspects without bail and a pandemic pause in gang-violence prevention programs.
In rural counties, where ties between police and locals are often less fraught, officials say the reasons for the rising violence are hard to pinpoint. They speculate that the breakdown of deeply rooted social connections that bind together many small communities, coupled with the stress of the pandemic, played a role.
Pastors point to the suspension of rituals such as in-person church services, town gatherings and everyday exchanges between neighbors. Such interactions can serve as guardrails, helping to prevent conflicts from turning violent. The psychological and financial stress due to isolation and job loss were especially pronounced in remote areas, where social services were limited even before Covid-19 struck, local leaders say.
As the pandemic took hold in the spring of 2020, fights between family members, acquaintances and even strangers escalated more frequently into deadly confrontations, authorities in some rural counties said.
-----
Obviously, causes could be different in different areas, but they could also be similar or the same in rural as well as urban areas.
But this seems to lend far more credibility to the idea that these trends have many different causes and that they don't fit nearly into any one box.
Sorry, I know that’s a cliché, but it became a cliché for a reason, and it definitely applies here. You literally have nothing but a correlation in time (and an appeal to your personal priors) and you’re using it to dismiss people who say the factors are complex. (And those people are not media editorial boards, as your phrasing implies. They’re named experts quoted in news reports.)
If we’re going to play the game of pointing at graphs and using our common sense to make something up, I think I can do better: The BLM protests and the murder spike shared a common cause – the widespread conviction that police didn’t care about the murder of black people and even did it themselves at the slightest provocation. After all, which would do more to make you feel you could get away with murder? Learning that police don’t care or learning that they won’t be driving around in their cars for a few months?
My raw speculation attributes the spike to a combination of high unemployment and stimulus cash at the time. Lack of a job gives plus a handful of cash led many people to venture into the business of buying and selling illicit drugs. But since there is no court to resolve disputes in that business, these new players found themselves on the wrong side of a gun.
Now the next subject of research should be whether the police in Uvalde were reluctant to act because of the BLM marches in their community in the previous year -- and they did take place -- and whether this isn't the question to ask if Latino police failed to act to suppress a Latino shooter. Still, it's important to research this issue as a school shooting is the most dramatic form of murder short of war and if there were any campaigns or policy changes or funding issues that could contribute to that phenomenon, we need to know it. No good saying that there were school shootings before the pandemic or BLM. THIS school shooting in Uvalde distinctly involved police unwilling to act forcefully, and blaming this on faulty comms or garbled reports when 45 minutes go by doesn't cut it.
I'm very confused by the emphasis of this essay. The narrative it presents is: police kill black people at a disproportionate rate; people protest against that; in response, police shirk their duty and oaths to keep people safe. If that's true, isn't that a *horrible indictment* of the police? In which case, why is it "politically inconvenient" for liberal media? And why would you say that the *protests* caused the homicide spike, rather than the police refusing to do their jobs?
Should a mixture of two ingredients cause an explosion, the explosion will follow the addition of the second ingredient, but that doesn't exonerate the first ingredient from causal role.
> and in the end focuses on rising gun sales (but guns are mostly bought by white people, and so can’t explain why the homicide spike was so overwhelmingly black).
Unless more guns in the hands of black is influenced by guns passing down from white legal purchasers to blacks getting guns through illegal trade. I'm not saying that's the explanation, just that it should be accounted for in your explanation of causality.
> I think there’s clear evidence that the current murder spike was caused primarily by the 2020 BLM protests.
Again, "caused" is problematic. You, yourself suggested the "cause" is cops doing less enforcement. Perhaps less enforcement is "caused" by the protests, or maybe it's "caused" by the murder of George Floyd? Or maybe it's "caused" myriad factors that contribute to hundreds of years of discrimination that led to Floyd's death?
How, along the causal mechanism, do you point to a singular "cause?"
> (Minor quibble – our philosophy isn't "Rationalism" – it's more accurately 'modern rationality' (no capitals) and it's less of a _compact_ 'philosophy' than a much more nebulous 'cloud' of ideas/aspects from all kinds of prior/current bodies of knowledge.)
- I appreciate you nuancing, though I can't say I understand the definition you are trying to express
> "I don't like that first part – you will _definitely_ "always find things to be critical of Scott"? Even if Scott states "I agree with Jordan."? That just seems _deliberately_ contrarian :)"
- I don't like it either because that's not what I intended to express lol. I meant it in the sense that people are different people, I know of no two people fully agreeing on every topic. I was ascertaining my right to disagreement, not my right to being a deliberate contrarian. As an aside, I do like stepping in other shoes for the sake of debate and taking on a contrarian position. The farther from my own beliefs, the more entertaining and thought provoking I find it.
1. I think not enough consideration has been put into the details of the mechanism by which police pullback supposedly caused more murders. It's possible, but how - exactly - did police work stoppages lead to more murders? I think that aspect of the hypothesis matters.
2. There's very clearly an uptick in Native American murders. Handwaving that away as statistical noise undermines the credibility of your argument, I think - it comes across as sweeping data that contradicts your hypothesis under the rug.
So, your analysis is that…. cops stop doing their jobs when publicly criticized? And *specifically* abandon their duty to serve and protect Black communities because they’re mad about and afraid of being critiqued?
I was inspired reading Scott Alexander's piece "What Caused The 2020 Homicide Spike?" So much so that I included it with 7 other primary resources proving there was a Ferguson Effect, where attacks on police and protest led to further bloodshed.
It's a long piece written the past 24 hours, but I excerpted the section where I focus on the Astral Codex Ten article from a couple weeks ago. Let me know what you think.
If you think murders increased from the “pandemic,” another hypothesis thrown in here, then you’re also off. Murders fell or remained static across Europe, Central America, and South America, and I’ve yet to find a single country that had a 2020 murder spike except the country (USA! USA!) that had thousands of “mostly peaceful” anti-police protests after George Floyd’s death, 7% of which turned violent. (Again, a situation where all the Minneapolis police involved were immediately arrested. So much for “no justice, no peace.”)
But critics say “more research is needed.” Which I’m all for, but somehow they missed the multitude of studies indicating a “Ferguson Effect” can be shown statistically and with strong methodologies. I don’t think that’s an accident that they ignore brave academics like Wilfred Reilly, Justin Nix, and Roland Fryer.
Is It Weeks of Months of Anti-Police Protest, Civic Disruption, and Attacks on Policing? (Bingo.)
Here’s a vastly more intelligent and thoughtful analysis by famous rationalist and blogger Scott Alexander. He also had a brouhaha with The New York Times, where he has legitimate beefs with them trying to out him publicly and tarnish his reputation, but I won’t get into it.
Simply, the creator of Astral Codex Ten is not beholden to media or academic pressures, where merely suggesting the “Ferguson Effect” means you’re not an “ally” and possibly “racist.” Alexander puts the near 5,000 single year increase in murders squarely on the BLM protests. Is there any incentive for career-minded academics or media journalists to promote this theory, even if painfully obvious? Of course not.
I think there’s clear evidence that the current murder spike was caused primarily by the 2020 BLM protests. The timing matches the protests well, and the pandemic poorly. The spike is concentrated in black communities and not in any of the other communities affected by the pandemic. It matches homicide spikes corresponding to other anti-police protests, most notably in the cities where those protests happened but to a lesser degree around the country. And the spike seems limited to the US, while other countries had basically stable murder rates over the same period.
I understand this is the opposite of what lots of other people are saying, but I think they are wrong.
Scott Alexander is brave, but he’s not alone. If more people stand up against the ideological grain, I honestly believe lives can be saved.
The timing made it pretty clear in most cases. I was looking through the data for Oakland, CA, and the spike was very large and happened almost overnight the week after the Floyd death. No other explanation could lead to such a quick shift like that. Pandemic, gun sales, poverty, change in the weather.....all other explanations would surely have more gradual increases and at different points in the year.
"I think there’s clear evidence that the current murder spike was caused primarily by the 2020 BLM protests...I understand this is the opposite of what everyone else says, but I think they are wrong."
You must not be reading many conservatives. Pretty much a standard gripe of theirs is that of course it was all BLM, and the media lies about it and obfuscates what's going just like they do with everything that they find ideologically inconvenient.
You make a good case, but I am left with the nagging feeling that you are affected by retrospective bias. It's not like in mid-March people said, "Pandemic! Argh! Kill! Kill!" At the time we all kind of thought we were talking about two weeks to flatten the curve. As weeks turned into months with no end in sight, lots more marginal behavior started, including the BLM riots themselves.
At least, I'd *like* to believe the BLM riots and both the overreaction and the underreaction to them were all ultimately caused by cabin fever, despair, and hopelessness, mixed with a certain amount of opportunism from some quarters.
I'm arguing from a sociohistorical point of view. I'm not going to say that evildoers get a pass because they were puppeted by the pandemic.
I’m confused. Doesn’t
1. The NYC shootings graph show that shootings were steadily going up during the month of May? I see the spike you mention in the caption but it seems to have actually started in mid-May rather than May 25.
2. The Minneapolis aggravated assaults graph counter your thesis? Assaults were rapidly increasing even before the protests started; they just continued their upward trend afterwards before coming down to current levels.
This is common knowledge in the Twin Cities.
I don’t consume a lot of right wing News but I do read the local newspaper. Mpls police were taking early retirement, going on disability or simply quitting in large numbers. Everyone in this area knows that the TC police are short staffed and trying to actively recruit. Those still on duty do feel unloved. It’s not just known by right wing media consumers but by everyone paying attention. Those people also know that the spike in murders is largely black on black violence.
It’s more complicated than what you are presenting here though. The scale of the riots was probably increased by the COVID lockdown. A lot of the young people that participated were available because things were locked down. The young were told they ran a less serious risk if they contracted Covid and they made up the majority of the civil unrest.
Then there was the timing of George Floyd’s death, right around Memorial Day. Weather is generally good at that time of year. If the death had occurred in February the reaction would have been less severe.
Why does an observation have to be right wing? Maybe all those mealy mouthed “it’s complicated” articles prove that the so called mainstream is in fact afraid to deal in reality. So much of reality seems to increasingly be relegated, by our self-appointed explainers, to this fantasy right-wing land. Meanwhile those supposedly in the mainstream know don’t seem to understand the depth of the groupthink they’re swimming in.
I agree that the BLM protests and follow-on effects are the most parsimonious explanation. That said:
> rising gun sales (but guns are mostly bought by white people, and so can’t explain why the homicide spike was so overwhelmingly black)
There were many articles indicating that gun sales spiked particularly for blacks. I have not done a deep dive, but just to suggest I'm not talking out of my ass:
https://www.cnn.com/2021/06/23/us/black-gun-owners-sales-rising/index.html
https://www.axios.com/2022/04/23/guns-firearms-people-of-color
https://www.theguardian.com/us-news/2021/apr/05/us-gun-ownership-black-americans-surge
From axios:
Retailer surveys conducted by NSSF showed that between 2019 and 2020, there was a...
58% increase in African Americans buying guns.
49% increase in Hispanic Americans buying guns.
43% increase in Asian Americans buying firearms.
"Although the George Floyd protests in May 2020 were the largest round of Black Lives Matters protests, there had been several previous rounds. Most notable were the shooting of Michael Brown in Ferguson, MO in August 2014, and the death of Freddie Gray in police custody in Baltimore in April 2015."
I got curious about the Baltimore/Freddie Gray numbers a while back and did a writeup.
For Baltimore there is NO DOUBT when the spike began. The cause is pretty obvious, too, though statistics can't tell us that.
http://mistybeach.com/mark/BaltimoreHomicides.html
There’s another possibility for the mechanism that you don’t mention. I know there’s been some academic literature on the idea that the community loses trust in police and places fewer calls to 9-11. I don’t know if this has good evidence for it or not.
"A priori there’s no reason to expect the pandemic to hit blacks much harder than every other ethnic group."
Really? There have been tons of articles about how the pandemic disproportionately affected black people. Yes, this comes from a media environment which is biased in many ways as you point out, but this claim seems quite plausible to me given that crises like this tend to have far worse effects on the economically underprivileged (as well as the possibility that there is sufficiently less trust in the medical system among African-Americans to have an effect as well).
Overall convincing article though I had matching priors previously. One nitpick:
Figure 1, weekly homicides per 100,000 city population appears to show local homicide minimums in March of 2017, 2018, 2019, and 2020, with local maximums in May of 2017, 2018, 2019, 2020. Is it possible that there is at the very least a simultaneous seasonal/weather effect? Based on that chart I can't help but think that if George Floyd died on March 1st attributed murders might have been somewhat lower, though there is likely some effect from the protests.
No other graphs in this article showed multi-year monthly murder rates so I have not referenced other data at this level of granularity yet.
I agree with almost everything in this post except for the media criticism parts. The conclusions seem very similar to this January New York Times article, for example:
https://www.nytimes.com/2022/01/18/briefing/crime-surge-homicides-us.html
My sense is that (1) most people believe the spike in murders was related to the Floyd protests, (2) most people believe that because the theory has been widely aired in the media, (3) the people and the media are almost certainly right about this.
What's much harder to say is exactly *how* the protests relate to the murder surge and what could we do about it?
1. Violent crimes rise during the summer. Following the lockdown, it's not surprising that crimes rose more.
2. I'd be interested to see deaths by economic status.
3. "but all things are smaller than other, larger things, and that doesn’t prevent them from being real or relevant." But it can mean that they are much, much less important.
4. It seems equally likely that the rises in murders are caused by the underlying factors which led to the protests. I won't insist on it, but it's at least as valid as Scott's "guesses".
The only thing wrong with this excellent article is the ninth word from the end, unless you were being ironic.
This government article talks about fewer people being put in prison.
And we know drugs were still being sold. Though the death rate from drug overdose was not as publicized during the pandemic, it was higher than before. Could fights have occurred more from drugs dealing, fights over money? Overcrowding in homes from lost jobs?
https://bjs.ojp.gov/content/pub/pdf/p20st.pdf
Steve Sailer discusses this often. One piece of evidence I've seen him mention is the increase in traffic fatalities, which might lend some evidence to the claim that it's a consequence of less policing. [1]
[1] https://www.takimag.com/article/the-racial-reckoning-on-the-roads/
I mentioned it in a comment, but you can also rule recent gun purchases out using trace data, as it takes years on average for a gun to go from purchase to crime.
Kind of feels like Scott came to his conclusion first and then is now reading the data to support that. But the data doesn’t at all look convincing to me. Clearly there was a spike around May, but the data shows it was starting in the middle of May? Floyd died on May 25, and while protests began the next day in Minneapolis, they really didn’t pick up steam across the country until a few days later. But, for example, the NYC chart shows a clear escalation that starts at the beginning of May.
Like Scott says the Minneapolis Aggravated Assault chart shows it’s obvious but all I see is that before Floyd died, cases were rising dramatically, hit their peak shortly after protests started, slightly dropped soon after and stayed at an elevated pre-Floyd rate the rest of the year.
The discussion of those NYT and Vox articles on the Ferguson Effect reminds me that so much of modern "reporting" and "fact-checking" is about determining who has the right or wrong vibes or context, rather than the facts.
https://twitter.com/NateSilver538/status/1541507378805440513?cxt=HHwWgoC8xeayxOQqAAAA
There was a point brought up earlier, regarding causal factors, and it seemed quite persuasive to me as a mechanism: if the black community, due to the protests (and the various grotesque murders that served to catalyze various BLM protest waves), began to perceive the police as less trustworthy, this would increase the homicide rate because homicide is correlated to distrust in institutions. It would also explain why the increase is largely confined to the black community.
So an increase in murders after the BLM protests is an indication of reduced policing. Because there is NO possible causal path between the BLM protests (or the events that provoked them) and increased murder, except for reduced policing.
And BLM and the pandemic are the only possible causes for an increase in murder in black communities, so if it doesn't look like the pandemic, it must be BLM.
You are better than that.
But what was causing the increase in murders? Just the reduction in police presence? I wish we had some data on the situation surrounding the murder. Was it drug gang violence, domestic violence, robbery, etc.
I suspect that attempts to claim the spike began in March in order to avoid political blame are specious, but I also have to question your implicit assumption that a spike due to covid would necessarily begin in March. What if it took a couple months for lockdowns to make everyone crazy? That could explain both the severity of the riots AND the murder spike.
Minor nitpick on the German murder graph - it looks like this is a graph of "Mord" (murder) as opposed to "Totschlag" (killing), which are separate offences in Germany. Shooting someone because you're angry with them, which is presumably what we're talking about with most of the murder spike here, would be Totschlag. Mord requires the crime to be evil in some sense (eg. treacherous, for the purpose of eating the corpse, committed with poison etc.).
The total rate for both in 2020 was 2,401, which was a 3.7% increase. (https://www.dw.com/en/germany-records-increase-in-murder-and-child-abuse-cases/a-57217668)
Unlike most people who are reading this piece and might bring it up, I would like to say that I think the new blue color for the site is nice and feels less generic and more intentional than the previous white background.
Minor gripe: I don't think the phrase "clear evidence" should be used in cases where there's been no attempt to actually isolate a causal effect (e.g. using a convincing differences-in-differences or instrumental variables approach, or an observational approach with a real claim to causality). The methodology this article uses could provide, *at best*, somewhat weak evidence.
Major gripes: making causal claims from time series data is really, really hard, and econometricians put a lot of effort into thinking about how to do it well. This article doesn't do any of that, and the genre "rationalist wades into ongoing academic debate without avoiding any of the pitfalls the field is already aware of" worries me epistemically.
Some specifics: many of your graphs don't show what you say that show, at least not clearly enough to count as evidence. A graph which is increasing before the event isn't evidence that the event had an effect, *even if you have a story to explain why the pretrend exists*. (It isn't evidence that the event had no effect either. It just means you can't make a convincing causal claim from it.) Similarly, the story you give to the Chicago police presence graph doesn't make that graph into evidence either -- you could come up with a similar-sounding story for pretty much any random walk.
To make an "it happened in place X but not Y" claim causal, you need evidence that X and Y would have otherwise responded the same to an outside shock. This doesn't work with America and any other country in crime (so many guns! no reason to assume the pandemic would have the same effect!), and also doesn't work with the black/other races response to the pandemic (black communities were hit much, much harder!).
It's methodologically very, very hard to distinguish "the protests caused this" from "the pandemic caused a huge increase but the first few months also had a large decrease because people were stuck inside" or "the pandemic caused a huge increase but only during the summer because that's when crime usually spikes" or something similar. Doing this right would be dozens of pages of econometrics, not a handful of graphs.
tl;dr there's causal evidence that policing reduces crime, but the spikes in this article would be much larger than any of the well-identified evidence. (I'd expect maybe a 10% increase at most?) I have no idea whether or not that's because they in fact very large, or because they're the results of other causes, but I don't think this article provides strong evidence that they're the result of the protests.
Do I get any kudos or credibility for having predicted the spike in print? Of course it was police pull-back, with possibly some community withdrawal.
https://quadrant.org.au/magazine/2020/09/and-now-we-wait-for-the-bodies-to-fall/
I'm surprised you think there's enough evidence here to confidently state that the pandemic had so little effect. Is your view that had the BLM protests occurred without the pandemic then we would have seen a similar homicide spike? Why can't both the protests and the pandemic be important factors? Why can't the pandemic have potentiated the protests and their knock-on effects?
You cite the timing of the homicide spike vs the beginning of the pandemic many times, but that's unconvincing to me. The beginning of the pandemic was characterized by widespread lockdowns and much less activity. It was also still pretty cold in many northern cities. As you said yourself, murder rates often spike with warmer weather--couldn't the pandemic plus warmer weather potentially explain the timing? (And yes ALSO the protests?)
I also don't think it's useful to compare the US with other countries--other countries have very different homicide rates to begin with (and different social safety nets, and had different responses to the pandemic, and different gun laws, etc). To say this proves the pandemic wasn't a factor in the US specifically seems questionable.
In general I think it's harder to disentangle the various factors here than you claim.
The more striking data is how wildly successful Mass Incarceration was at reducing murder (and likely other) crime from 1994-2012
Another data point: Miami neither had widespread riots or a major defund the police movement. (Or at least if there was one it didn't succeed.) There were peaceful protests but there was minimal rioting. Miami does have a significant Black population too. Violent crime decreased while arrests went up representing (at least according to the local politicians) the police interrupting people trying to commit crimes. This is despite not having as much of a weather change or sticking people indoors as long and having pretty widespread gun ownership and a long history of organized crime. (Seriously, it's Miami. It's still a major drug port.)
Anecdotally I lived near a lot with a lot of high end cars. The police basically had some officers watching it full time and they'd arrest a few attempted car jackers a day during the first few weeks of the protests. After a week or two the thieves (but not the protests) died down. Either due to thieves taking the hint or literally removing them from the streets. If there was a pullback I didn't notice it. If anything it seemed like the police were more active.
Someone should make a list of "uncomfortable facts that the right/left media don't want you to know, that are actually backed by solid evidence".
As one of the people who questioned your attribution in the previous article, this post causes me to take a significant update in the direction of your thesis. I'm still not totally convinced, primarily because causal inference on observational data is incredibly hard, but the evidence you present here is better than evidence I've seen presented in favor of other points of view.
I also think I was more importantly wrong on a meta point. Part of the reason I questioned your assertion is because I thought something like "this is just the sort of thumb in the eye of sanctimonious establishment media that would be catnip for Scott", so I thought it was likely that you would be biased on the subject and may be making overstated claims on questionable data. While I still think you may have some bias here, I also think your conclusion is honestly arrived at with proper consideration, and I regret any implication of unvirtuousness in my previous comment.
I predict a spike in comments elsewhere in what used to be called the blogosphere to the effect that "once again Scott Alexander has shown his true colors and good people everywhere will denounce him and never read him again."
I appreciate your analysis here, but there were numerous points that left me rather confused. First, "A priori there’s no reason to expect the pandemic to hit blacks much harder than every other ethnic group" — no? the lowest income groups in the US are blacks, Native Americans, and latinos. All of these had an uptick in the homocide rate during the pandemic (e.g. chart you present midway down). An entirely plausible alternative hypothesis is that lockdowns pushed people in low income groups into more dire economic situations, and this lead to more murders. It seems all non-white/non-asian groups had a rise in homocide rates that you're ignoring here, and class is an obvious potential confounder.
Second, a huge problem in the "Effects of other Protests" section is that protests are perfectly confounded with the underlying event perceived as a huge social injustice worth of mass protest. There are many possible causal hypotheses here, two that come to mind are:
(1) death by police → protest about policing → policy of reduced policing → more homocide/crime
(2) protest about policing ← death by police → less trust in social and government institutions, less prosocial behavior → more homocide/crime
Increase in homocides could be (1) alone, or (2)'s right path alone, or (2)'s left and right paths, or (2)'s left and right paths, with the left path also going along (1)'s entire chain to also lead to more homocide (it's hard to draw DAGs here, so hopefully this makes sense). The point is that death by police and protest are coupled, so it is nearly impossible to disentangle "the effect of protests" from "the effect of less trust in social institutions less prosocial behavior because of police murder". So I think the claim "If Black Lives Matters protests can cause homicide spikes, we would expect to see one around this time also" isn't alone sufficient, because it fails to control for the upstream cause of the protests. In reality, I bet it's both — protests reduce policing, reduced policing changes the economics of crime a bit, and less trust in our social institutions leads to more antisocial behavior. But I wouldn't be so quick to blame protests as much as what caused the protests. Protests could act primarily through an interaction in that they spread awareness of police murder and that reduces social trust much more quickly and widely than without protest.
Regardless, I do agree it is rather unfortunate the media covers these issues with such heavy partisan lenses, and not as a really tricky social problem with unknown causes that we should really work hard to understand, so we can try to have some idea of a sensible intervention.
The charts you show reinforce my distaste for rolling average plots. They often confuse when the effect actually happens. That's particularly a problem when you're trying to tease out causation, which is often what those plots are used for!
If you think your data has some artificial noise or periodicity, try leaving it in and just labeling it! I think you'll find that most people can figure that out.
I think a factor that's been overlooked is fiscal stimulus contributing to gray market demand for firearms. I've been through the legal system twice (once during and once 5 years before pandemic), and this time it def seemed like the judges were having a strong reaction specifically to the prevalence of guns at the time. Imo, everyone having more money at the same time really raises the perceived value of having a gun (either for greed or protection), and this could explain why there aren't similar spikes in countries where guns aren't readily available
> But there are lots of reasons to expect that the Black Lives Matter protests would cause police to pull back from black communities in particular.
Here's data on arrests by race: https://crime-data-explorer.app.cloud.gov/pages/explorer/crime/arrest
2019 -> 2020
Whites: 5,236,036 -> 3,771,245 (28% decrease)
Blacks: 2,063,765 -> 1,413,290 (32% decrease)
I'm not following how depolicing leads to an immediate uptick in homicide. Petty crime, sure; robbery maybe; murder seems like a stretch. I agree impunity rates affect criminal behavior *on some time frame* but is the feedback that immediate? An alternative hypothesis is that the depolicing is a response to the uptick in crime caused by calls for black lawlessness combined with demoralization of policing. I would guess that the type of policing that falls the most is the kind of "circling the block in a police car type" of beat, not the difficult detective work and DA buy-in needed to bring someone in for homicide. (I know this firsthand: my brother was murdered in August 2020, and we're still fighting to get his killer arrested nearly two years later.) So I think Scott needs to go into more detail on the mechanism behind depolicing -> more murders. But I enjoyed the piece, and I agree that the NYT sucks on this.
Great analysis. I think events turned the social tenor toward chaos, and people who are only marginally civil exploited the opportunity. Why are certain races victimized, certain properties targeted? Some blame the pandemic, global warming, colonialism, police abuse and murder of criminal suspects, racism, inflation, white supremacy, etc. If one is determined to act unmoored and/or violent, to dishonor the social contract, any excuse will do. Someone usually benefits from chaos.
The response time of the homicide effect to the protest "intervention" seems awfully fast - faster than I remember the politics moving. When did police departments actually start quitting or pulling back? One could imagine a more immediate effect of "the police are dealing with the protests and not other things", if your theory is that police presence is strongly preventative of crimes. In sports you often double-cover your opponent's star player, so #2 or #3 ends up getting more plays. Maybe it's as simple as the cops having steady-state capacity but not surge capacity.
But at least for homicides the picture still doesn't quite make sense. Even though the timing looks somewhat compelling, the motive is still muddy. Why does inter-racial grievance lead to a spike in intra-racial violence? Does police presence actually work? Are cooler heads who would ordinarily prevent this violence out at the protest? Are people deliberately taking advantage of the lapse in policing to go gunning for someone? Are people of their own race just close at hand when they get back from the protest with their blood still running hot? This seems like a critical part of the story to nail down for it to emerge into that respectable conversation and out of the fringe right.
"The murder rate in the UK fell in 2020-21 compared to the previous year:"
This is strong evidence *against* the theory that BLM protests caused the homicide spike, because there were massive BLM protests in the UK. All sorts of statues were torn down. Wikipedia's map of protests in the UK shows an incredibly high density of protests. https://en.wikipedia.org/wiki/George_Floyd_protests_in_the_United_Kingdom#/map/0 And while black people only make up 3% of the UK's population, they made up 13% of London's population, quite close to their % of the US population. And homicides *fell* in London in 2020. https://www.onlondon.co.uk/the-number-of-homicides-in-london-fell-in-2020-what-explains-why-these-sad-statistics-change/
This is the big problem with the argument that "COVID was a global event, but the homicide spike only happened in America." The George Floyd/BLM protests were also a global event! There were protests on every continent including Antarctica. https://en.wikipedia.org/wiki/List_of_George_Floyd_protests_outside_the_United_States A bunch of protests outside the US got quite rowdy. Just to give one example, on June 6, 2020 the US embassy in Mexico City had to be locked down due to violent anti-police brutality protests in the vicinity that saw cars being set on fire and businesses being vandalized. https://www.newsweek.com/us-embassy-mexico-city-locked-down-due-george-floyd-protests-1509112 Yet I've seen other people cite Mexico's *lack* of a homicide increase in 2020 as evidence for the BLM protests causing the homicide increase.
And as other people have pointed out, several of your graphs show an increase starting before May 25, not after it. Here's another graph that shows the same thing, scraped from the Gun Violence Archive that supposedly covers all of the firearm homicides in the country: https://pbs.twimg.com/media/E20sKyAVEAAOAMk?format=jpg&name=4096x4096 It clearly shows an increase above 2019 levels starting in April, not at the end of May. The source is this blog post by none other than Steve Sailer, who claims it was sent in by a reader, so it clearly meets the Criterion of Embarrassment. https://www.takimag.com/article/the-racial-reckonings-new-normal-50-murders-per-day/
You might say that that still doesn't show an increase starting shortly after the pandemic. But we also have to take into account other effects of the pandemic. The most obvious is people spending a lot less time outdoors. Here's a graph of mobility in 2020 compared to 2019: https://pbs.twimg.com/media/FIxre9dVEAQKqP4?format=jpg&name=medium It shows a massive decline starting in March, bottoming out in April, then returning to 2019 levels...in the middle of May. People are obviously less likely to commit murders when they spend less time outside, so this data massively confounds any time series data on homicide.
There are also other time series graphs that I think are worth a look. Here's a graph of opioid overdose deaths, which clearly shows a significant increase starting in February 2020. https://pbs.twimg.com/media/FIxrSCuUcAADqZc?format=png&name=900x900 Alcohol sales show a very similar increase, which strongly suggests that the former data was due to an increase in consumption, not hospitals having less resources or anything like that. https://pbs.twimg.com/media/FIxq-bgVcAA5jKr?format=jpg&name=medium
We would expect that more consumption of alcohol would lead to more alcohol-related murders. We would expect that more consumption of illegal drugs would lead to more drug dealers shooting each other. We would expect decreases in mobility to offset this in the period immediately after the pandemic started. We would expect this to hit black people harder than other races because black people are more likely to deal drugs (and probably even more likely to be involved in urban drug gangs that are more likely to get into violent territorial disputes with each other than IE meth-dealing biker gangs). It isn't that hard to think of reasons why this phenomenon might hit homicide numbers harder in America than in other countries. So it seems extremely plausible that at least some of the increase was due to changes in the consumption patterns of intoxicating substances, and at that point why not apply Occam's Razor?
I admit that one flaw in the alcohol and drugs theory is that Mexico didn't see a homicide increase in 2020, but it's possible that COVID border crossing restrictions made drug smuggling more difficult and that offset some of the impact on the Mexican drug scene.
I also think this article presents a pretty compelling argument that the late 2010s homicide increase may have also been due to drugs (in that case, due to the opioid epidemic) and the timing may have been just a coincidence. https://www.vox.com/policy-and-politics/2018/2/6/16934054/opioid-epidemic-murder-violent-crime
And here's one thing that I have yet to get a good answer for from "Ferguson Effect" advocates: why wasn't there a homicide increase in Los Angeles following the 1992 riot, which was worse than anything that any single American city has experienced since then? Here's a graph of homicides in major American cities in the 90s: https://pbs.twimg.com/media/FQxUdYVVUAImPk6?format=png&name=medium Los Angeles doesn't stand out at all.
Heritage has nothing to gain by using a less provokative title. The left have overwhelmingly demonstrated their hyperpartisanship on this issue and categorically are not interested in the truth on this issue. A strong claim, but when you have the peak "respectable" news site in the NYT basically reporting something they don't believe to be true, it shouldn't be a controversial claim.
From the book review:
> Murders definitely rose a little after Boudin took office, but that’s because that was also when the Black Lives Matter protests happened, which demoralized police and led to a so-far-permanent spike in murders nationwide.
That specific phrase, "demoralized police" was something that I took issue with, and made a comment on. The specific claim was not that police pulled back, it was that they were demoralized. And now in this article you say:
> I don’t want to speculate on which of these factors was most decisive, only to say that at least one of them must be true, and that police did in fact pull back.
But in the previous article you directly attributes it to demoralization, which was what I took issue with, and why I left a comment there.
The evidence does seem to support the case that the protests (and related unrest, especially in black neighborhoods) led to an increase in the murder rate. Though from the stats here I'm not convinced that reduced policing is mediating this? If a policing pullback were the cause, I'd expect a murder spike earlier march/april when policing was reduced for the pandemic response. The Freddie Gray graph also shows a more gradual decline in arrests starting well before the protests, whereas the spike in murder rate was relatively abrupt and more tightly coupled to the protests.
I think Scott makes a good case for the pullback in policing that followed the BLM protests contributing to the 2020 homicide spike. But there's another factor that may have played at least as large, and I think even larger role: the social and economic stress caused by Covid.
One good measure of this additional stress is the number of overdose deaths. This from the CDC:
"Provisional data from CDC’s National Center for Health Statistics indicate that there were an estimated 100,306 drug overdose deaths in the United States during 12-month period ending in April 2021, an increase of 28.5% from the 78,056 deaths during the same period the year before."
One can reasonably infer from this, and the increase in overall drug use it implies, that the emotional impact of Covid on the residents of low-income, inner-city areas, most of them living in relatively cramped conditions and with very few financial resources, must have been huge. The added stress falling upon a population with very easy access to guns, suffering acute financial distress and unprecedented social pressure, could have led to an increase in homicides that was unrelated to police withdrawal.
The timing of the spike in killings, a few months after the Covid lockdowns began, is at least as consistent with the evidence as the explosion in BLM street activism, which itself might have been, in part anyway, a consequence of the unparalleled intensity of social stressors in 2020-21.
This needs further analysis.
“The idea that the police have retreated under siege will not go away. But even if it's true, is it necessarily bad?”
I couldn’t stop laughing at this
Would benefit from more discussion of 2021 and 2022. Here in New Orleans the murder rate is even higher this year, which it’s hard for me to plausibly attribute to BLM protests in the summer of 2020. Also the “pandemic started months earlier” point is unconvincing, could be that it’s prolonged isolation that causes breakdowns in relations
I live in Columbus Ohio. A city not in these charts. The city population is about 900,000 and the Metro is about 2.1 million. In the city of Columbus, murders did increase dramatically in 2020 and the city set a record with 175 murders. Whether or not it tracked the timing in other areas, I do not know. Whatever the reason, the spike continued and got worse in 2021 when there were 205 murders.
Just when you thought you knew what was going on, and for equally mysterious reasons, there have been only 62 murders as of June 25, 2022, the 176th day of the year. At that rate the number of murders will drop to about 130 a 37% decrease.
There have been no changes in the political leadership of the city since the beginning of 2020.
Here is a link to a Columbus murder database:
https://www.dispatch.com/in-depth/news/crime/2021/07/21/columbus-ohio-homicides-map-and-database/7812208002/
It may be paywalled.
Police responses are almost certainly one mediating factor, but let's not overlook the simplest possibility: BLM implied that police all across America murder blacks; this made a lot of people really angry; angrier people are more likely to kill people. I suspect the Marxist founders of BLM would consider a spike in homicide a feature, not a bug. It both signifies an increase in revolutionary consciousness, and destabilizes the society to be overthrown.
This isn't a tight empirical/causal argument so the level of confidence in these statements exceed the available evidence. For example, what if the pandemic caused BLM and both had contributions to homicide. The level of data storytelling presented here could be explained by this same model and we're back at square one.
It's clear that the media has an agenda - and perhaps they left the conclusion to this unusually and purposefully vague. However, I think that vagueness is the right answer to reach, however inconsistent, and "data" presented here is short of being the evidence needed to match the confidence of your claims.
I think the issue of guns is kind of too easily hand waved here.
The idea that, because most guns in the US are bought by white people they can't have an impact on a subcommunity seems specious.
For example, people buying weaponry to smuggle down into Mexico is not a large proportion of US gun sales. But, if you then said, therefore the availability of guns from the US is not a factor in making the Mexican cartel wars super violent, it would be wrong.
Similar with all those countries that aren't awash in guns that didn't see a spike in the murder rate.
I like the article and I find it mostly convincing as a whole.
One part which I don't understand is the relationship between people choosing to commit more murders and the decline in police presence/action/screening or perhaps it is the anger stirred up by the protests and angry people kill more often, just like hot people do with more summer murders. Are we just x amount of irritation away from attacking each other to death?
I get random stops by police might cause an arrest for illegal firearm possession or something which then prevents a murder by removing the gun and interrupting a crime of passion....but can that really add up to so many more murders?
I get we look at the protests and we look at the police actions...but no one is saying it is the police or protesters themselves going around murdering people. It is random other people killing more.
Do people who want to kill think they are less likely to be caught and decide to kill more often? Are we saying that police presence and actions to arrest for minor crimes are a huge deterrent and/or they interrupt such rare events by harassing the general population? Or a targeted population of young men of a certain race? It just seems like you could stop thousands of people and not interrupt or stop a single murder.
One factor which is hard to pin down is if the desire or circumstances for murder to occur have themselves gone up. Which I don't understand. The police killed a black guy one didn't know, so one is going to go out and also murder some other guy?
I don't get the logic...I saw only 3 police patrols instead of 7 police patrols in my neighbourhood over the past week....time to go kill someone?! I don't get it. Is seeing a few cops and a handful of extra guns or drugs taken off the streets here and there really the reason people choose to kill others or not? It isn't like murder isn't already very very illegal with heavy penalties. Or that a lower police patrol rate means you think the cops are going to work less hard in a murder investigation? Do killers think this?
Is a 10% or 20% less likelihood of feeling like you'll get caught the threshold for people to go out and commit the murders they already wanted to commit? Is that all that separates life from death for so many victims of murder? Are gangs just going nuts because there might be a 30% reduced rate of cops driving around?
I just feel like something is missing here. Do people want to kill more than before? Do they want to kill but feel like they were being held back and those barriers are gone? Is killing like other crime from the 'broken windows' hypothesis and simply needs a context? Killing seems like far less of a crime of opportunity than shoplifting does and context feels very weak as an explanation. But again, summer murders are higher, so maybe context matters far more than I'm thinking?
I suppose people just wanted to murder each other more in 2020. Why do americans like to murder so much more than other nations. Perhaps there are tipping points or thresholds of poverty or population or gun ownership or gang violence or drugs and despair or undemocratic abusive workplaces, but even when you 'control' using statistics for the number of guns or whatever...the US murder rates are higher. And that's not even counting the invasions, wars, or drone strikes of the american empire!
For weird reasons I've read a lot about this issue so here are a few points I want to raise wrt to your argument:
- Using only city data seems like it could be misleading if you're making conclusions about a murder spike that happened nationally. CDC data is probably more useful here if you're interested in figuring out the timing of the spike. The CDC found that murder was already up 20% nationwide in April 2020 and up almost 30% for blacks. Then there was an additional spike in June after the protests started.
- The causality of the protests on the murder spike is kinda dubious even though part of the spike does line up with the protests beginning. I brought this up in a previous post but the 2020 BLM protests saw an outpouring of young people, particularly black and in urban areas, into the streets after months of lockdown. When tons of young people start all coming out into public at once, you're going to get a rise in murders no matter whether police behavior changes or not. For example holiday weekends in the summer (i.e. the 4th of July) are often the deadliest times of the year for murder because people are socializing all over the place and you end up with tons of drive-by shootings and the like. So based on this you could explain the spike by saying that propensity to murder was already up before the protests (which I think is born out in the CDC murder stats) but lockdowns were holding people back a bit from actually carrying out murders - until, that is, the protests hit and the dam of social distancing broke.
- Interestingly murder went way up in 2020 but almost all other crimes didn't. Both police reports and victimization surveys (which don't rely on changes in police reporting) corroborate this. So this begs the question of why, if de-policing caused a murder surge, it didn't cause a surge in other crimes as well. Is murder just unusually sensitive to changes in policing, compared to other crimes? It's possible, but the difference would have to be huge to explain the lack of any changes in, for instance, theft and robbery.
Shockingly, validating crime increases it. Shocking.
Somehow this reminds me of a tweet thread I saw recently presenting a study that paid Fox News viewers to watch CNN instead. The author made it out to be about the pitfalls of partisan media in general—they just happened to pick fox because it matched the current administration. But pretty clearly the big selling point was that it confirmed liberals’ biases about right wing media without requiring them to face up to any of their own media’s blind spots.
Wonder what a study would look like that paid NYT readers to follow ACX for a while.
https://twitter.com/dbroockman/status/1510607409294692352?s=21&t=1eem7vmYLYmQtvOBatR2_A
Has anyone bothered to see if all-causes deaths went up by the amount indicated by the increase in reported murders? Meaning, did more people die in those communities because there were more murders, or were there the same number of deaths and more were officially classified as murders?
Because, maybe I'm getting a biased view from listening to liberal-minded podcasts and interviews and articles on this subject, but it really feels to me like official crime statistics have more to do with what the police are incentivized to report than with the actual amount of crime happening. And they were certainly incentivized to report an increase in black-on-black crime immediately following the protests.
It look like the spike is something like a 50% increase immediately, dropping back to baseline over a few months. Do I think police have enough wiggle room in how they report things to invent such a spike in their statistics even with zero change in actual rates? I think it's plausible, yes - that seems like an increase of 1 homicide per 100k citizens per month, not a huge number of bodies to reclassify.
Classify some marginal things as homicide instead of manslaughter or accidental death. Does the coroner usually take 2 months to file cause of death paperwork? Get them to dump the backlog this week and speed up the process this month. Do you often ignore people who look like gang members or criminals when they try to report a crime? Take the report this time.
I feel like this wouldn't be hard to do, especially if your previous incentive was to keep the reported numbers as low as possible to look good to your superiors, so you're starting from an artificially depressed baseline.
I'm not claiming to be an expert or know for sure that this is happening or could happen, but from everything I've heard about police departments being driven by arrest metrics and community-safety metrics and how this has affected reporting over the years, I'm surprised this isn't one of the first hypotheses tested for whenever the crime numbers do something odd.
And of course the framing of this conclusion as "right wing" is a political construct in itself - assuming that suburban whites won't care why policing declined because they're so fearful and easy to manipulate (having been a suburban white surrounded by suburban whites my whole life, I'm not sure they're wrong).
David Simon, for instance, just did a whole mini-series about how, yes crime has gone up due to police protest, and how that makes police a bunch of cowards who don't deserve the public trust.
You can ideologue with any data, and I won't even be mad. The annoying bit is the strategizing. I don't care what you think, just say that thing and not the thing you think you need to say to make me think what you think.
This made me wonder about homicide rates after the LA riots. This Time Mag article points out that immediately afterwards “In May and June, only four people were killed, down from twenty-two during the same period in 1991. Drive-by shootings fell by nearly 50 percent from 1991 to 1992, and gang-related homicides by 62 percent.” The decrease is primarily attributed to the effects of the formal truce and peace pact agreement between the Crips and the Bloods the day before the riots. So this essentially supports your hypothesis as well.
https://time.com/6049185/los-angeles-rodney-king-misunderstand-what-happened/
In “Votes, Drugs and Violence,” Trejo & Ley make the argument that in a society with some corruption (overlap of bandits with govt&police), periods of political upheaval generate ripples through the bandit-establishment interface. This triggers turf wars among bandits, leading to increased killings. It can be as direct as “the leader of the corrupt police was promoted & left the city, so now other corrupt police and their factions in the bandit community are battling for control.) In order to have a society with fewer bandit killings, the corruption has to reach a steady state with leadership hierarchies that are clear to those involved.
Is this the mechanism by which changes in protest behavior cause increased murders? I don’t know how tightly the police-bandit connection is held in these cities.
I do wonder if, rather than A caused B, perhaps both A and B are caused by a different C, not the pandemic. Perhaps whoever organizes protests made friends among the bandits and conducted a two-pronged campaign, political street protests to weaken police plus community violence to weaken the community & establish dominance over underground economies.
The fact that you need this many graphs to demonstrate this to smart readers is the reason I'm giving up on smart people. They are worse than dumb people.
Need pages of sources to tell you that there's a correlation between a spike in the murder rate, and the most expensive riot in US history--which happened at the same time, occurred because of the police, coincided with a widely-broadcasted pullback in policing, and of which oodles of videos of the reality were available. Gunfire going off like popcorn in all directions.
Also need pages and pages of sources to tell you that women are not the same as men.
Need pages of sources to prove anything that an effete intellectual would find unpleasant if it were true.
Well written, I applaud the effort to rebut griping from commenters. Thank you for researching and putting all of this together.
Police do useful work *and* commit atrocities. Probably the same is true of the justice system in general. What policies would help maintain or increase the useful work while making atrocities less likely?
It seems to me like there's a great deal of evidence here that around the time of George Floyd's murder, and a couple other high-profile police killings, there was immediately after a spike in homicides.
But it doesn't seem like any of the above data proves, or even really provides strong evidence for the idea that, it was because of the specific mechanisms you state.
Here is a way to test the theory - do homicides increase after killings *of police*? I would think that actual cops being killed is a bigger cause of cops being afraid to do their job, then BLM protesters.
Deaths of police on the job went way up in 2020, but IIRC the increase was due to cops dying of COVID.
Perhaps respectable people should read more right wing sources then.
Thank you Scott. So much enlightened (also from the comments)
Unrelated to the content of the post, but I much preferred the plain white background to the new pale blue background. Just my two graphic design cents.
A couple of reasons for thinking covid played a sizable role in the U.S.:
When I was a kid watching Westerns, the term "masked man" was synonymous with bandit. I suspect that encouraging masked men to roam the streets contributed to the return of car-jacking during the pandemic.
Also, lots of dangerous people were released from lock-ups during the lock-downs due to fears of infections in jails and the like. Of course, this was also something the Establishment had been wanting to do since the war on mass incarceration got going about a decade ago, so it's hard to disentangle covid and wokeness.
Similarly, both covid and wokeness played a role in the war on college admissions testing: in-person testing was called off in 2020, and then many colleges announced they were getting rid of testing, to one extent or another, to fight racism. I call this the Not So Great Reset in which temporary expedients during the worse of the pandemic feed into long-term changes in directions the Woke had been wanting to go.
Portland had 4 homicides in the first 6 months of 2020 and 55 in the last 6 months. The only thing that changed from the first half of the year to the second was that BLM and Antifa became very "active" in the city so I think we can all guess why the homicide spike occurred
I understand THAT this is supposed to be some sort of right-wing point, but I don't understand WHY without some SERIOUSLY uncharitable assumptions about how the right-wingers making the point think about black people. After all, the idea of a police strike, by which police are punishing communities for their insolence by failing to properly protect them from violent crime, would seem to me to be a pretty LEFT-WING point about toxic police culture.
On a tangential note, a few weeks ago my wife called the police because there was an African man sitting in a crosswalk drinking hard liquor at about two in the afternoon. The police arrived, helped him up, and took him aside to politely ask him if he was doing OK, how much he'd had to drink, and whether he needed medical help. The man was obviously somewhat indignant, feeling racially profiled, but my wife was not the first person to call the police, because he was sitting IN A CROSSWALK and was clearly not doing great. At no point was anyone in danger of being arrested, jailed, shot, strangled, or hurt in any way, because this was in Germany, where police are generally well trained and disposed to be helpful and resolve situations as peacefully as possible. This is policing that is actually useful.
> Here’s number of arrests in Chicago. We can see that it goes way down in March when the pandemic starts and everybody (including police and criminals) are indoors, then starts going up again before the protests. Then after the protests it goes back down and stays down. My interpretation is that people complied with the strict lockdown early in the pandemic, that effect was played out by May, and then separately the protests caused a longer-term decrease in policing.
I found this unconvincing. On all other graphs the effect of the protests was immediate, but here there was seemingly a several week delay? On top of that, the downturn in arrests coincides with a downturn in the average for previous years too, suggesting that this could be a normal thing in June for some reason. (Though usually with a much smaller magnitude.)
The Minneapolis traffic stops graph also shows the downturn clearly starting before the protests, and it doesn't look like data smoothing since there's a small spike shortly before. I found it strange to see this graph presented as though it backed up your conclusion.
There's another explanation, which might be slightly conspirational:
Police officiers and prosecutors have some pretty hefty leeway in what gets investigated and prosecuted as a crime.
During a time of threat to the police budget, it is in their interest to make the world appear as "unsafe" as possible.
It doesn't mean they're inventing murder cases, but simply that their behavior changed compared to the previous baseline.
It would be hard to prove, maybe a dip in manslaughter, suicides and accidental deaths?
Another point, the assault clearly started rising before the protests. Random noise?
"I think there’s clear evidence that the current murder spike was caused primarily by the 2020 BLM protests."
I'd argue that causation went like this:
The riots start -->
The Establishment more or less encourages them as "mostly peaceful protests" -->
The police are discouraged and potential bad actors (e.g., shooters and reckless drivers) are encouraged.
Potential bad actors are encouraged in two ways: they fear the cops less so they do things like carry illegal handguns more and speed more. And they feel more exuberant because Society is telling them they are the good guys and the cops are the bad guys.
Case and Deaton coined the term "deaths of despair" for white working class losers killing themselves, dying of opioid addictions, or drinking themselves to death. The high rates of black homicide and traffic fatalities deaths after May 25, 2020 look more like "deaths of exuberance." For example, mass shootings (4 or more killed or wounded) at black social events went way up in the summer of 2020. Traffic accident deaths among blacks were 36% higher in the 7 months after George Floyd's death in 2020 than in the same period in 2019, whereas the other 87% of the population saw traffic deaths go up 9%.
Didn't a lot of cops quit as well in relation to the protests? Leading to fewer police and more experienced cops replaced by less experienced cops. I think I have read that a few times.
In 2019, Dean Baquet, chief editor of the New York Times, addressed his staff after Robert Mueller's testimony. He said that while he, too, was sad that the NYT's Plan A to get rid of Trump via RussiaGate had failed, they should not despair because the NYT was rolling out its Plan B to get rid of Trump via playing up racism.
My impression is that, for months, many Democrats saw the "mostly peaceful protests" as a way to get rid of Trump, either in November or possibly earlier in a mostly democratic color revolution. Thus, Joe Biden, who is not by nature a progressive extremist, didn't seem to recognize that the riots were bad for his campaign until the third day of the Kenosha riots in later August 2020, when he finally started calling for calm.
Interesting that the correlation is parsed as "BLM causes homicide spike" and not "video footage of police murdering a man causes homicide spike".
"The 2020 homicide spike primarily targeted blacks"
As an Englishman, it's always a surprise to hear about the depth of the systemic discrimination in America. It's not just people and institutions, you also seem to have racist graphs!
Everybody already knew that the murder rates are driven by the minority communities.
The light blue background is much nicer than the bright white we had before.
“Several people in the comments questioned my attribution, saying that they’d read news articles saying the homicide spike was because of the pandemic, or that nobody knew what was causing the spike.”
I’ve read many “news articles” which variously asserted that the spike in violent crime and the decrease in ratings for TV sports were due to the pandemic. If someone writes an alleged news article saying that, you know that he’s a transparent, pathological liar. If someone cites a transparent, pathological liar as an authority, what does that make him? An “expert”? A “tenured professor”? A close relative of mine?
Time-series data on homicide rates across all European countries can be found at Eurostar. This table in particular.
https://ec.europa.eu/eurostat/databrowser/view/tps00146/default/table?lang=en
Reassuring to find an independent voice on a topic that has become (needlessly) controversial. If people are being killed, when it happens is a good start in figuring out why.
I'll put forward another theory to add to the mix. Blacks in the United States are less restrained in their behavior because they feel like they are winning. Winners get to do what they like. Similarly, fans of sports teams tend to riot more after a win than after a loss.
Has anyone worked out the total number of above trend deaths for the year? Would help to know how many extra individuals died in absolute terms. Quick estimate looking at the first graph seems to be a 1.4% increase / 100 000 people x 330million citizens = ~4620 individuals, more accurate calculations welcome
This is a strange post by Scott.
It spends a whole lot of words conclusively proving that there is a strong correlation between the BLM protests and an increased homicide rate.
Then it triumphantly declares clear proof of causation.
And, as far as I can tell, there is no actual case for causation. The arrow could go the other way, both could be caused by a third or multiple other effects, or there could be circular causation.
My takeaway is that Scott is fed up with what he perceives as a liberal propensity to refuse to believe inconvenient things, and that led him to beliefs further than the evidence supports.
Explanation for the German murder-spike in 2018: The German Police adds murders to the statistic in the year they learn about them, not in the year the murder happened (since otherwise you would have to correct the statistics when it takes them a while to figure out a murder happened). In 2018 there was a trial against a man, Niels Högel, who confessed to around 100 murders he commited by killing patients while working as a nurse. This alone accounts for half the spike you see in the statistic.
While I agree with the research and the numbers, I don't think your conclusion is well formulated. Saying that "the current murder spike was caused primarily by the 2020 BLM protests" makes it sound like there is a one-step causal connection. As in the act of protesting bad police work will always cause more homicides. This kinda reads like blaming the protesters for the murders, as if protesting somehow leads to murdering someone. This can't possibly be a satisfying conclusion if your goal is to know why more homicides happen directly after BLM protests. Of course this is fantastic if all you care about is finding SOME truth to prove main stream media wrong.
You go on to mention that "My specific claim is that the protests caused police to do less policing in predominantly black areas." just to fully ignore it the conclusion, as if the police simply have no agency in this matter. What you really do here is no better or worse than NYT or Vox since all you end up doing is focusing on an irrelevant truth to strengthen your contrarian position. Anyway, thanks for linking protesters to homicides, surely nobody would use that to justify crushing legitimate important protests.
Another potentially interesting factor to consider:
Most courts closed for the pandemic. Many of them, especially in cities, didn't reopen for several years or reopened with much lower capacity. A lot fewer murder cases were being tried in court because of the pandemic. The justice system either had to detain people without speedy trials, or it had to not prosecute crimes. Neither of which are good options.
If this were the main cause of the increase in murders, I would expect a gradual increase starting in March. Instead, we see a sudden increase starting in May. So I don't think that this is the main contributing factor. But it could contribute to why murders haven't gone back down.
Maybe I missed this in the article -- is there evidence that the depolicing caused the increase in murders (rather than just both being responses to the Floyd incident)? Genuine question. In my head, I had assumed that it was something like increased anger/tension/despair in black communities as a response to Floyd's death.
If I believed the Intercept graph, the safest city in the US was Baltimore and, um, I think the rest of the data quoted contradicts that.
Also, how did Portland have *negative* murder rates? Was there a necromancer passing through who resurrected a lot of the formerly dead? There is definitely something odd going on with that graph, and I think it may have to do with "percentage change in murders" - percent change compared to when? If there were 1,000 murders in March and only 100 in April, that would give a negative percent, but it wouldn't mean that there were no murders going on.
And if I instead believe those lying liars in Oregonian media, Portland had its own little sharp increase in murders over the period 2019-2021:
36 murders in 2019
55 murders in 2020 (83% increase according to the FBI as reported by KOIN 6 News, with nice colourful bar charts and what the hell is going on in Milwaukee, I would not have pegged it for 'murder spree city'? https://www.koin.com/news/crime/fbi-data-portland-homicides-up-83-from-2019-to-2020/)
90 murders in 2021 and the fun goes on:
https://www.opb.org/article/2022/01/15/2021-was-a-record-year-for-homicides-in-portland/
"City police and officials say last year’s increase — which disproportionally impacted Portland’s Black community — was fueled by gang-related arguments, drug deals gone array and disputes among people living on the streets. In addition, the situation was exacerbated by the pandemic, economic hardships and mental health crises.
In a May statement, Portland Mayor Ted Wheeler warned residents that “groups” were traveling to the city from Washington and California to “engage in and advance gun violence.” In addition, Wheeler said “groups involved in this violence have issued an order to shoot someone in the next 30 days or be shot for not showing loyalty.”
Similar to the ’90s, Portland officials have said the city faces a rampant gang problem. Comparatively, police and residents say the boldness of shooters, their young age and the amount of shots fired surpass what they have seen before."
What I want to know is, why aren't black people vocally concerned about the rise in homicide in their communities? Could it just be that the media-favored activists don't care, but the average black person does?
Not just homicides. Motor vehicle accident deaths increased a great deal, especially amongst blacks. I plotted car accident deaths and homicides on a monthly basis from 1999-2021 as standard deviations of each groups' monthly mean to help separate seasonal effects from secular trends. It's pretty compelling. You can see both increased in 2014 and then again a great deal around May 2020.
https://twitter.com/RCAFDM/status/1536419294107602944
I'd like to add one small factor: Drug trafficking is way up.
When you heard about a 40% increase in overdoses, did you consider the size of the iceberg underneath that? Anybody who ever had a habit started again during the lockdowns. With lots of demand, lots of product and lots of manpower, local dealers have been making $$$. Naturally, there has been some conflict up the chain, and anywhere there is that much $$$, there will be criminal violence.
But that is maybe 15% of the increase in crime. The police pullback is the rest. Based on conversations I have heard *IN JAIL*, amongst criminals and Sherrifs, criminals know the cops' hands are tied, they are in defensive mode. Criminals are far more aggressive from the start when the feeling on the street changes from,
"violence or shootings are certain to bring a big LEO response",
to
"Cops on duty are down 25% already, and some are at mandatory diversity seminars. 95% chance we just get away with this crime right now...." and
"...any fool has a 90% chance of getting away with murder."
This creates some real sympathy for the cops, in a soldier-ly way.
BR
Looking at all of this data, the more interesting question to me now is related to the figure of US murder rate by year and asking what caused the murder rate to double in the decade following ~1964?
The background turns blue, and then Scott backs the blue. Coincidence?
View from the inside. I'm two steps removed from several police officers, and the reduction in policing verifiably comes from these factors:
1) Cops willfully and consciously stopped going to black neighborhoods. Literally stopped responding to 911 calls there, because the risk of themselves getting roasted in the media, skewered by a social media mob, fired, and jailed was not worth the risk of going to the black area. They stated this privately, and they acted on this on the ground. A great example was the entirely justifiable Wendy's Atlanta shooting. The police officer was acquitted, but the media mob attacks on him literally caused his familial relatives to be fired from unrelated jobs.
2) Cops quit their jobs. Staffing across the country is down, especially in urban police forces that had to deal with ACAB social pressures. They decided to work lower stress jobs such as private security.
It was less of a "own the libs" thing as it was a "well if this is what they want I guess we'll give it to them" thing and a "they've made the personal risks too high to do my ordinary job" thing.
It seems to me that Americans have a lot of trust in punishment. Not all Americans, of course, but I see this weird hypnosis where the idea of punishing someone leads to believing that the punishment will have exactly the effects one wants. The unwanted behavior will go away, and there will be no blowback or side effects.
This shows up both in long prison sentences and in the left wanting to punish the police. I'm not saying punishment never works, but it's not all that reliable.
https://www.statista.com/statistics/585152/people-shot-to-death-by-us-police-by-race/
Note that more white people numerically (though not proportionally) are shot by police than black. I've been angry that police shootings of white people (some of them sketchy) aren't taken seriously, but maybe the net effect has been better for white people. On the other hand, the justice system shouldn't just be about net effect.
One more angle, the police are trained to be very afraid of the public-- as long as that's in play, it's going to be hard to get them to treat people more carefully.
> A priori there’s no reason to expect lockdowns and “cabin fever” to hit blacks much harder than every other ethnic group.
Blacks are more likely to be poor, and so live in tinier and more rundown apartments and in neighbourhoods with fewer parks, and they have more children. Sounds like a recipe for cabin fever to me. That would also explain the uptick in arrests prior to the George Floyd murder.
> No country except the United States had a large homicide spike in 2020, which suggests that the spike was unrelated to the pandemic and more associated with US-specific factors, for which the BLM protests and subsequent pullback of policing in black communities seem to me to be the most obvious suspect.
To play devil's advocate, the US arguably has the strongest culture around freedom to move and associate.
What about 2021 and 2022? As far as I can tell the trend continues. An overarching theory as to perhaps why:
https://www.bloomberg.com/opinion/articles/2021-02-26/why-is-the-u-s-murder-rate-spiking
This raises the question: what's the "natural" level of black-on-black crime? Will it always rise as policing intensity decreases?
Wouldn’t you expect a crime spike after the protest even if BLM caused no increased propensity toward crime?
State #1: Lockdown. People are inside so there are fewer potential murder victims. Crime is down.
State #2: Everyone is outside again, and total number of murders go up.
Protest certainly seemed like the de facto end of lockdown where I live.
Owen G made a similar point in a previous comments thread.
This does not invalidate your other points, but it could mean that we should take the spike in late May less seriously.
Thank you for this. Maybe you are the brave, investigative journalists I've been dreaming of? I have one question, it's a doozy and no one is bothering because the corporate media are stasi hacks. HOW MUCH DID THE "PEACEFUL" PROTESTS OF 2020 REALLY COST? One measly Axios report, estimating the damage between $1 and 2$ billion, covered only two weeks- when we all know it went on much longer. Some of these areas will never recover- and they aren't in the rich parts of town.
I don't think the police decided to relax policing in black neighborhoods as much as they were ordered to do so by people now trying to rewrite history and hide their actions.
Looking at fig. 1 - eyeballing the data, you could just as easily say that there was no spike before july 2020. The peak in may is only very slightly higher than peaks in the previous 3 years. I would have liked to see a statistical analysis of whether this peak sufficiently explained by random noice. Certainly, if you consider the last 20 years to be homogenous (probably not a perfect assumtion, but still, looking at the last 10 years seem just as arbitrary) that spike does not seem that abnormal at all.
I guess I'm arguing that the whole post may be about nothing - sorry....
It's obvious the BLM protests and riots caused the homicide spike, but it is also obvious that the pandemic caused the BLM protests and riots to happen on a scale ten times or more greater than they would have otherwise. Perhaps without the pandemic we don't get widespread BLM riots and an increase in homicide.
What does "the current murder spike was caused primarily by the 2020 BLM protests." actually mean?
If social issues had been accumulating like tinder, and May 2020 sparked the match, does that make the match the cause? If the match is the cause, then you should be able to decrease murder rates by targeting the cause... but I laugh at the idea that BLM Protests is the social issue to address in order to reduce murder rates. It is the underlying social issues that are driving these murder rates and need to be addressed. This is why I'd like to understand what it means to scott claim BLM as the primary cause. This article lacks a broader context to make the conversation meaningful or practical. The context feels like "here is my defense of my prior claim". Cool?
I see many faults with this analysis. We lack a comprehensive baseline for the seasonality of murder rates year over year. There is no explicit causation, just casual inference. I agree with the inferences but not the conclusion. And the opposing perspecties are dismissed outright.... Scott is explicitly trying to explain his perspective so this post is inherently biased rather than exploratory.
Do I think the civil events of 2020 Summer affected murder rates? Absolutely... and it happened during a pandemic at the end of Trump's presidency. Still, it feels completely irrational to try and focus on a singular primary cause to a complex social issue. Again, if "BLM Protests" were the match, how much tinder was already there and what is the real cause? What is the purpose of writing this? What is the message to be taken from it?
I was discussing this essay online and my interlocutor pointed out, and I'm copying/pasting this:
"A very simple falsification of his theory would be if the bulk of the increase was in-household homicide."
Do we have numbers for that?
(To be honest, I think that if the bulk of the increase was in-household then that opens up even uglier questions but, hey. The point of science is to go where the data inexorably leads.)
I think your interpretation of the effect of the moving average is backwards.
When using a moving average, the measurement should lag behind drivers, not go ahead of them.
Stylised example: If the murder rate is always 5/day, then on Day N the atmospheric lead concentration goes up and stays up and changes the murder rate to 10/day on all future days, then the 7-day moving average murder rate will start rising slowly on Day N+1, reach 10/day on Day N+7, then stay there.
But you use the fact it's a 7-day moving average in some of the charts to argue for the measurement starting to go up *before* the event, as far as I can see, and that's completely backwards. Moving averages can't give us advanced predictions of the things that will eventually cause them to have risen!
All round, a lot of the very good time-resolution graphs seem to show the effect starting before the protests, and even before the protested murder itself (which was May 25th, I think), more like the beginning of May. That's very confusing to me.
But it's always tough to say when a trend change actually occurs in noisy data. Oftentimes if you just looked up to Day Y, you'd say, well, yes, Day Y is high and it's been trending up since Day X, but it "trends" up and back down again all the time just due to noise, and there's nothing particularly "trendy" about the Day X to Day Y period. But the if you look onwards up to Day Z, you find out that the increase just keeps on going and going until Day Z is clearly way out of noise range and there's definitely been a start of an upward trend at some point, and then the temptation is to go all the way back to that last inflection point on Day X and call that the start of the new upward trend, even though when you'd only seen up to Day Y, Day X didn't look at all special, just one more inflection point in a noisy bounce up and down.
You discount the spike in gun sales too much. Sure the protests led to the increase in gun sales so they are correlated. Homicide rates are still high though and that is because gun sales are still high. Police do tend to pull back when convenient to set a narrative and in some countries you even see that with gang violence.
"Guns are mostly bought by white people, and so can’t explain why the homicide spike was so overwhelmingly black"
I think Scott's overall analysis is right, but this particular sentence doesn't ring true, because legally-bought guns have a tendency to make their way into the black market via theft. In 2014 Tennessee allowed residents to keep guns in their vehicles without a permit, and then in 2020 there was a huge spike in gun purchases, related to the general craziness of 2020. Here's a graph of "guns stolen from vehicles" in Memphis, TN, by year.
https://content.govdelivery.com/attachments/fancy_images/TNMEMPHIS/2022/06/5997697/guns-stolen_original.png
It was possible to predict it for exactly these reasons. I did on June 2, 2020: https://twitter.com/LechMazur/status/1267863723013820416.
I'm really skeptical about one connection here. I don't think the homicide spike could be mediated by cop pullback. The timing doesn't work. Intuitively, there should be some amount of lag behind any of the reasons you proposed for cop pullback. Funding isn't cut instantly. The 'Defund the police' messaging took a little while to dominate. Police didn't instantly grow resentful; most condemned the Floyd killing anyhow, so only became resentful due to the protests. And this fits with the graph of arrests in Chicago; the arrests only started to fall off a week after the protests started.
But the homicide spike happened as soon as the protests happened. It needs to have been mediated by a much faster effect. I'd suggest the mundane 'there were just more people spending more time outside.' We know that murders in cities already go up when the weather is nice and people are outside, and go down in the winter and in rainstorms/blizzards. It makes sense that if some social effect made even more people go outside, there might be even more murders.
Eyeballing these graphs and invoking priors about crime does not make a "compelling statistical case". Correlations in timing are weak evidence for claims that one thing was "caused primarily" by another, not "clear evidence". The observations made in "right-wing sources", or at least those you repeat in this post, are also made in the other sources you link. I don't believe it's correct to say a point has "gone unnoticed" when it is taken into account in justifiably more careful causal inference.
You've shown through your use of statistics that BLM protests directly correlate to the rise in homicides in black communities in the cities where the protests took place. Great. Now how did BLM protests influence this rise? Working out this more difficult question that should produce a nuanced answer, but you'd going to need more than stats to make it clear. What you've done at this stage is give fodder to those who care to use it.
Really liking the new design here.
well done
After notable incidents of police brutality, calls to 911 drop. Most interestingly, some cities have microphones designed to pickup gun shots and pinpoint them to a block. This provides useful data since the microphones will pick up all gun shots and you can look at call logs to see how often neighbors call the police when guns are fired.
The problem with the claims of the Ferguson effect is quite frankly cops do not prevent many murders. If you want to kill someone, you're not going to do it if you see a cop standing on the street, but that just means you'll do it when your target isn't near a cop. If you kill someone because you are in a fit of rage, again that often happens in places where you normally wouldn't have a cop just standing around anyway.
The Ferguson effect is not that protests cause crime spikes, it's that when trust between police and a community breaks, crime spikes because the community no longer trusts the police and will not engage them, for example, if they hear two people arguing in the street late at night. My father-in-law was Italian and grew up in Newark NJ and his motto from the streets was "you never call the cops unless someone has a knife sticking out of them"
Now I also think 2020 had another impact on crime. The pandemic hurt open air drug sales and as a result dealers tried to move in on each other's territories to maintain themselves in a shrinking market. At least from watching a lot of shooting headlines from NYC in 2020 and early 21, a lot of them seemed very much like hits rather than the peak of a generalized crime wave (i.e. someone zips up on someone in scooter, shoots him, zips away). This can be tested when there's good data on all types of crimes. Usually one would expect homicide to be on the top of a crime pyramid with lesser crimes making up larger numbers underneath. In that model if homicides increase, all other crimes should too. If, however, we find most other crime stays the same but homicides go up much more, that indicates something is going on with the distribution of crime.
Scott: When you consider murders exclusively your conclusions make sense, but when you consider the stats for other types of violent crimes, it makes less sense. If policing went down post BLM, how do explain the fact that other types of violent crime didn't rise along with the murder rates? Violent crimes other than murders have only risen slightly (too bad I can't paste graphs into my comments).
Another data point you missed is that gun ownership among blacks went up 58% in 2021, and it had been rising for several years before that.
OTOH gun ownership also went up significantly in Latino and Asian communities (albeit not as fast as in black communities). I'd be interested if we saw any increase in homicides and or crime in either of those communities — which you'd expect to see if police started patrolling less post BLM.
I don't have an alternative theory, but any explanation needs to explain why other other violent crimes haven't risen alongside murders.
"I think there’s clear evidence that the current murder spike was caused primarily by the 2020 BLM protests."
Very sound analysis, but the conclusion seems flawed. The murder spike was likely caused by the police pullback from the streets. While the police may have pulled back in response to the protests, I do not see how we could say that the protests 'caused' the police pullback (and hence the murder spike). If I call Scott a bad logician (which he is not) and he pulls a gun on me, is it right to say that I caused him to shoot me?
It feels a bit weird to attribute it to the protests and not the killing that they were a reaction to.
I could have saved this author a lot of work. What causes lawlessness? Leadership practicing lawlessness as an example to the people.
Tired of lawlessness? Stop electing Democrats (or helping them cheat).
Somewhat off-topic, but do we know why murder rates increased rapidly from1964 - 1974, and then decreased rapidly from 1990 - 2000?
"I looked into Central America, and found that Honduras, Nicaragua, El Salvador, and Guatemala also all saw murders decline rather than spike in 2020."
A gigantic portion of crime in Mexico and Central America is tied to smuggling drugs into the United States. In 2020, the US imposed restrictions on entry over land from Mexico. https://www.federalregister.gov/documents/2020/03/24/2020-06253/notification-of-temporary-travel-restrictions-applicable-to-land-ports-of-entry-and-ferries-service One would expect this to make drug smuggling across the US border more difficult, and therefore to lead to a decrease in the volume of the drug trade in countries South of it.
But drug overdose data indicates an increase in demand for drugs in the US during the pandemic. https://pbs.twimg.com/media/FIxrSCuUcAADqZc?format=png&name=900x900
Taking these two factors into account, it isn't hard to see why this might lead to a large increase in homicide specifically in the United States. Demand for drugs increased, but dealers who sourced their products from South of the border were unable to meet that demand. So more money went to dealers who got their drugs from other sources, and some of them used their increased resources to expand into territories of the dealers who got their product from down South. When people started to go outside again in Summer 2020, some of those new territorial disputes turned violent.
It makes perfect sense why we would see this more in the US than in other countries. Compared to other developed countries, an unusually large portion of American illegal drugs come across one particular land border, the American drug trade is unusually violent, the easy availability of guns might have something to do with it, etc. That's without even getting to the question of whether cultural or other idiosyncratic factors might have led to a greater increase in drugs and alcohol consumption in America during the pandemic than in other countries (for the record, the first Google hit for drug overdose deaths in the UK in 2020 shows a pretty small increase of just 3.8% https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/bulletins/deathsrelatedtodrugpoisoninginenglandandwales/2020 ). And it makes perfect sense why the homicide increase would be greater in demographics that are more likely to be members of drug gangs in high population density areas where we would expect territorial disputes to be more frequent.
It all seems very consistent with the available data.
I wrote a whole substack article about how the media has its own agenda, and since I'm a big believer in self-promotion, I'll link it here.
https://questioner.substack.com/p/the-censored-of-censored
But in all seriousness, I think that the refusal of so-called "expert sources" and media outlets to self-correct when they get something wrong (like their explanation for this big crime spike) is a huge problem. It seems to me like it's mostly based on ego. People in power don't have the humility to admit when they're wrong and because they hold power, they typically don't have to.
That's why you should vote for me when I eventually run as president. I promise that if any media source or person of influence makes a mistake and doesn't own up to it when new information comes out, I will punish them in a terrifying way. Egotistical narcissism is a hard personality trait to eliminate but when it is put in direct opposition to the self-preservation instinct, anything is possible.
The left doesn't look good here and they haven't for a long time. Dating back to The New Jim Crowe (2011?) I've seen a lot of slight of hand from the left, and then when someone raises an inconvenient fact, there is more smoke, mirrors, and ad hominem arguments.
The causal chain between BLM riots and the rise in violent crime seems crystal clear to anyone without an agenda. Considering the progressive effort that goes into convincing us that black white disparities are the mostly the result of a history of institutional racism (a claim that is highly disputed even among many black intellectuals), this link looks much more direct and well supported by evidence. If we can't agree on something so seemingly obvious as criminals do more crime when they know they won't be punished for it, how can we solve any meaningful problem? Isn't this a "hard conversation?"
This highlights an area that cause a lot of compassionate but logical people to question whether the left cares more about their own sacred ideas rather than the underlying problem.
"In the second graph, you can see a more traditional presentation of homicide rates, which shows them *shooting up* after Gray’s death to a level higher than they had been in the previous twenty years."
Insert laughing facepalm emoji here.
There's an alternative explanation that fits the evidence here: the killing of Floyd itself caused the crime increase by damaging trust in the police, which led to an increase in retaliatory violence.
https://www.city-journal.org/retaliatory-gang-violence
I posted some of this as a reply but it occurred to me that it might be worth elaborating on as a main comment. The police arbitrate violent disputes. If your friend is shot and the police to deal with it, there's probably no further violence: the perpetrator is arrested and sent to prison, end of story. But if you don't trust the police to deal with it - because there has been an extremely prominent example of them being untrustworthy - you might decide to take matters into your own hands and seek violent revenge.
This is just as compatible with the timing of the evidence you've described, including the pre-pandemic stuff in Baltimore and Ferguson.
This argument could be strengthened or weakened by research on how trust in the police changed among young black men, especially those adjacent to or involved in gangs.
I did see another comment mentioning that reported trust of the police among white people was affected more but at the risk of speculating a bit, even if a white American reported trusting the police less post-Floyd, I'm not sure they would turn to revenge violence in the event that they were the victim of a crime. Given that retaliatory violence is already common in gang disputes, I think it makes sense that it would go up there and not suddenly appear in other parts of society.
This old Tabarrok papers cleanly explains a mechanism through which we should expect riots to lead to permanently higher crime rates. Having read it long time ago, when I saw these riots I was pretty confident that they would lead to permanent increase in crime. As a bonus its explanation is less morally charged than your proposed mechanisms.
The thing is, given constant police resources the equilibrium crime rate is not self correcting. Exogenous temporary shocks to total crime will lead to a permanent change in total crime.
This happens because the decision to commit crime depends on the share of crime that is caught, but with constant police resources the share of crime that is caught depends on total crime.
So what happens is a temporary exogenous rise in total crime leads to share of crime caught falling, this makes people more prone to commit crimes, and that makes crime rise even more. When the exogenous shock is gone, there are still more crimes than before, so the share of crime caught is still lower, perpetuating the new higher crime equilibrium.
(Incidentally that is a minor plot point in True Dectective's second season, where a group of mostly law enforcement officers take advantage of the LA riots, and the consequent straining of police resources to make a hit on a jewelry store)
https://link.springer.com/article/10.1007/BF02298409
An interesting question would be how the homicide rate moved in cities that previously experienced the Ferguson effect compared to large cities in general.
Baltimore, for example, saw a spike in homicides after the Freddy Gray riots in 2015, which has remained until now. Notably, unlike the rest of the country, it did not have a homicide spike in 2020. Total homicides in Baltimore actually dropped in 2020.
This fits nicely with the thesis of this article. Baltimore has experienced the effects of the effects of the hypothesized depolicing, the George Floyd riots thus had little additional impact.
Edits: 1 ) removed grammar mistake, 2) clarified in the first sentence that I'm objecting mostly vehemently to Scott's characterization of the criminological literature regarding the OG Ferguson effect (see his section titled "Effects Of Previous Protests") which serves as kind of a robustness check to his claim that protests -> homicides.
Scott, you are wrong (about the original Ferguson effect, which serves as a robustness check of your more general claim of protests -> homicides)! And I can show you why. Why should you read this random comment among a sea of meshugas? Argument from authority: I'm a PhD candidate in one of the top criminology programs. In other words, I'm intimately familiar with both the content area and the issues of time-series analyses. I'm happy to privately verify this and/or share full text articles I source below if they are paywalled. Key points:
0) Begging the question and also not citing relevant literature that tests+disproves your claims
Throughout this letter (which should be revised to become a mea culpa), you conflate two separate hypotheses:
a) There is a POSITIVE ferguson effect (aka protest effect) on DEPOLICING (aka less 'active' or 'proactive' policing)
b) There is a POSITIVE ferguson/protest effect on violent crime, PARTICULARLY HOMICIDE
Hypothesis A appears true: police do pull back from policing when facing public scrutiny. The extent of this pullback is debated. For more info, see MacDonald 2019; Marier and Fridell 2020; Mourtgos, Adams, and Nix 2022; Rosenfeld and Wallman 2019. However, you beg the question by showing hypothesis A is true, and claiming hypothesis B is also true!
Hypothesis B is plausible and worth exploring! This is because at the extremes both 1) more and/or well targeted police(ing) reduces crime (Evans and Owens 2007; Sherman 2022) and 2) poorly targeted and/or no police(ing) increases crime (Loeffler and Braga 2022; Piza and Conneally 2022). Deterrence - particularly versus nothing - does exist (Nagin 2013)!
However, the problem is that we have actually tested hypothesis B and found it lacking! See the Criminology and Public Policy Volume 18 Issue 1 (2019), particularly Rosenfeld and Wallman 2019. The issue is that there's an assumption that arrests (and other CJS interventions) have a UNIFORM marginal effect, but this may not be the case. Instead, imagine that the marginal effects of incrementing/decrementing a(n) arrest/imprisonment/search/stop etc. is conditional on the cumulative exposure, much like a laffer curve! In this case, it is not so clear that reducing police interventions would result in increased crime, as long as the reduction isn't 'too large'. See Owens 2019 for a very good short summary on this economic perspective on CJS interventions.
So, TL;DR hypothesis A doesn't mean hypothesis B, and in fact we (criminologists) have already done this work and found hypothesis B lacking.
1) Causal and validity problems - aka time series analysis is hard
You've done some work, but it isn't even close to enough. To help get there, read the very approachable McCleary, McDowall, and Bartos 2017, with particular attention to the chapter on construct validity.
2) too high an expectation of there to be a parsimonious explanation - aka homicide is a highly contingent social event
It shouldn't be surprising that a reasonable desire for a parsimonious explanation - like protests -> homicide - falls apart. Why? Because homicide is actually the culmination of highly complex social phenomena. You have to have 2+ people, a place, a social context, a particular reason, and (usually in the US) a gun! Parsimonious explanations seldom provide enough of those elements.
3) Why does it matter if you are wrong?
Well, because being wrong about attributing the rise in homicide constitutes perpetuates a historic trend of perversely individualizes general social problems, creating variations of the following viewpoint (not to mention Sailer in the comments of course): 'The reason black people suffer in the US is that the poor and/or black community is composed of individuals who have low self-control, make poor decisions, have broken families, bad cultures, excessive disobedience, immoral norms, and self-perpetuating poverty. Therefore poor and/or black people profligately inflict crime (particularly bad ones) both within their community and across the US.'
Basically, this view of the world is like watching a cut of The Wire that only features drug busts and scenes of violence (aka the media up until 5-10 years ago). This means that structural conditions are ignored, or perversely endorsed in a 'race realism' mode to morally justify inequity. This is a lazy interpretation and it's uninspiring you stan for it when, in other arenas, you seek novel explanations. Like, instead of this hogwash you could write about when regulations and higher education intersect with negative externalities (e.g. Lovenheim and Owens 2014).
Bibliography:
Evans, William N., and Emily G. Owens. 2007. “COPS and Crime.” Journal of Public Economics 91(1):181–201. doi: 10.1016/j.jpubeco.2006.05.014.
Loeffler, Charles E., and Anthony A. Braga. 2022. “Estimating the Effects of Shrinking the Criminal Justice System on Criminal Recidivism.” Criminology & Public Policy (online first). doi: 10.1111/1745-9133.12588.
MacDonald, John M. 2019. “De-Policing as a Consequence of the so-Called ‘Ferguson Effect.’” Criminology & Public Policy 18(1):47–49. doi: 10.1111/1745-9133.12430.
Marier, Christopher J., and Lorie A. Fridell. 2020. “Demonstrations, Demoralization, and de-Policing.” Criminology & Public Policy 19(3):693–719. doi: 10.1111/1745-9133.12492.
McCleary, Richard, David McDowall, and Bradley Bartos. 2017. Design and Analysis of Time Series Experiments. New York, NY: Oxford University Press.
Mourtgos, Scott M., Ian T. Adams, and Justin Nix. 2022. “Elevated Police Turnover Following the Summer of George Floyd Protests: A Synthetic Control Study.” Criminology & Public Policy 21(1):9–33. doi: 10.1111/1745-9133.12556.
Nagin, Daniel S. 2013. “Deterrence: A Review of the Evidence by a Criminologist for Economists.” Annual Review of Economics 5(1):83–105. doi: 10.1146/annurev-economics-072412-131310.
Ouss, Aurélie. 2020. “Misaligned Incentives and the Scale of Incarceration in the United States.” Journal of Public Economics 191:104285. doi: 10.1016/j.jpubeco.2020.104285.
Piza, Eric L., and Nathan T. Connealy. 2022. “The Effect of the Seattle Police-Free CHOP Zone on Crime: A Microsynthetic Control Evaluation.” Criminology & Public Policy 21(1):35–58. doi: 10.1111/1745-9133.12570.
Rosenfeld, Richard, and Joel Wallman. 2019. “Did De-Policing Cause the Increase in Homicide Rates?” Criminology & Public Policy 18(1):51–75. doi: 10.1111/1745-9133.12414.
Sherman, Lawrence W. 2022. “Goldilocks and the Three ‘Ts’: Targeting, Testing, and Tracking for ‘Just Right’ Democratic Policing.” Criminology & Public Policy 21(1):175–96. doi: 10.1111/1745-9133.12578.
Gell-Mann Amnesia runs both ways. I've found claims about what "The Media" says to nearly always be false or cherrypicked when I bother to check them. (Also, is The Economist not part of "The Media"? If it is, then citing their graphs here is a self-defeating argument.)
To throw in another theory for the general situation, maybe double time for overtime is a bad incentive. Perhaps the police are too sleep-deprived to be thinking well. I've never seen discussion of when dubious behavior happens compared to number of hours worked.
People wanting to party after bars closed, restaurants. No fun. A riot was a big party where anything goes. For that crowd, police officers would be in danger. Gangs could kill them. All the prisoners being released would likely go find those who got them incarcerated. Deadly game for officers, families, informants. All people who needed to make a living suddenly hitting the streets. Many may have died from drug overdoses.
I'm inclined to think the decrease in policing was a large factor---and indeed there's a large literature showing that increasing policing decreases crime---but it does seem like there are some bugs in the argument here.
How does this comport with the observation that all kinds of bad behavior went up substantially? E.g., https://www.slowboring.com/p/all-kinds-of-bad-behavior-is-on-the
It seems implausible that BLM protests would cause a rise in people to get into fights on airplanes. Given that we know anti-social behaviour generally went way up, the simplest explanation for the murder rise is that it's the right-tail of the general phenomena. The concentration of rise in murders in black communities could be explained by a model that just says everyone now commits crimes 1 degree more severe than they were pre-covid.
That leaves the questions of why the murder phenomena is America specific, and why the violence spike started in May. One possible answer: CARES act stimulus checks were sent out in April. This large cash infusion may have enabled many relatively poor people to buy (additional) guns.
(Both for and against this theory, this paper claims that gun purchasing started going up in March, but finds no association between increased gun purchasing and increased violence at the state level https://injepijournal.biomedcentral.com/articles/10.1186/s40621-021-00339-5)
My point here is not that this is a better theory for the murder spike, but just that it fits with the main points of your argument and is additionally consistent with the general rise in (non-murder) anti-social behavior.
They let offenders out of jail and prison that normally would have been in cages. Most offenders are black. They went home and killed people. Very simple.
It was unemployment. Peaks of unemployment correlate with peaks of homicide. There are other long term trends that raise or dampen the height of the peak, but unemployment seems to me an obvious explanation.
Here's my rough sketch of the causal mechanism:
Some say 5-10% of young men (mainly) are strong, impulsive, angry and tough and not very skilled. Many times they are marginally employed. You get covid and this group gets thrown out of work en masse.
What do they do? They get into all kinds of trouble. They get into fights with one another. Some of these lead to homicides. Mystery solved.
Here's the unemployment chart since 1960. Notice how the peaks overlap with the murder chart for the years prior to 1992?
Then after 1992, when the great unleaded gas effect washed over, the peaks in homicides became *much, much* smaller, but still correlated with unemployment.
However the peak in unemployment in 2020 was off the charts, and so was the blip up in homicides.
https://data.bls.gov/generated_files/graphics/latest_numbers_LNU04000000_1947_2022_Annual%2BData_data.gif
Increases in police retirements and difficulties in recruiting tends to support the points made regarding police reactions to Ferguson, Grey and Floyd effects. What is worth recking you retirement, your life, your assets or even criminal prosecution? At some point you have to be practical.
First, what is this "elite establishment" that you refer to? The WSJ, the NYT, or USA Today, or all three or are you referring to news networks like Fox or Yahoo News?
Your calling a "protest" a "riot" already highlights a bias. Protests can become violent, but riots start out that way--I should know, I worked in a prison and went through several riots and I can testify that I would never call them protests, but you seem to have no problem calling a protest against police violence on black people a riot. Is it because that's all black people can do? And was it "your city" being looted? That sounds like something a white person would say while people of color facepalm in the background--I saw the same thing with Occupy where suburban white kinds would march in some inner city neighborhood they didn't know jack about while shouting "Whose streets? Our streets!"
Sorry, I'm not intentionally trying to be confrontational, but I thought that this site was more about using rational means to address problems.
The argument here is that BLM protests caused changes in police behaviour which in turn led to the homicide spike, while covid and restrictions did not in turn lead to the same.
BLM caused the spike, covid not. Timelines don't match up otherwise. So far pretty convincing!
At the same time, though, we can see that covid restrictions did lead to less policing, just like BLM, without being correlated with any homicide spike. Wouldn't this imply that BLM caused the spike by some other mechanism, or at least that there is some meaningful additional variable making a difference in how police withdrawal affected homicide rates between the spring and summer?
It's not totally off tye wall to use the weather to explain the increased severity of the protests/riots. The February Revolution happened because St Petersburg had a "warm" spell in the middle of a general workers' strike, and at the same time as International Women's Day. If the weather had stayed cold in February 1917, the protests would likely have been less severe and the Tsar may have held on a few months longer. If Floyd had died in February or March 2020, maybe the protests are less intense.
But the weather itself can't explain the sustained levels of violence. 2020 didn't see a spike on Memorial Day followed by a decline to pre-2020 levels. Murder rates went up and stayed there. And murder rates have retained their seasonal pattern (higher in summer than winter), which is almost-but-not-quite counter-cyclical to covid and covid restrictions. If the spike in violence was simply an outburst of suppressed energy from lockdowns, then lockdowns would decrease the murder rates and ending the lockdowns would cause a temporary uptick. That isn't really seen in any of these cities.
On the comparison to other countries - I can attest to the strictness and insanity of Covid rules in Germany, as I spent the entirety of the pandemic there with occasional trips back to the USA. Lockdowns in Germany were far stricter and longer lasting, and letting up lockdowns didn't lead to mass numbers of shootings or stabbings. It did lead to a lot of outdoor celebrations with masked people dancing 2 meters apart from each other, but that was about it.
"Police interpreted the protests as a demand for less policing, and complied."
Coming from a family of New York City policemen, this reason caused great laughter around the table. The police do not comply with the demands of protesters but The second choice also seems wrong
"Police felt angry and disrespected after the protests, and decided to police less in order to show everybody how much they needed them."
Yes to 'angry and disrespected', but the decision by the police was to punish, to avenge, and in doing so to violate their oaths.
I did a search on the murder spike in this period in the NYT:
“ Changes in policing. The fallout from the 2020 racial justice protests and riots could have contributed to the murder spike. Police officers, scared of being caught in the next viral video, may have pulled back on proactive anti-violence practices. More of the public lost confidence in the police, possibly reducing the kind of cooperation needed to prevent murders. In extreme circumstances, the lack of confidence in the police could have led some people to take the law into their own hands — in acts of street or vigilante violence.
The timing supports this theory, with homicides rising unusually quickly shortly after George Floyd’s murder and the ensuing protests. Killings also spiked in 2015 and 2016, after protests over policing during those years.”
https://www.nytimes.com/2022/01/18/briefing/crime-surge-homicides-us.html?referringSource=articleShare
Another NYT article
“ The protests that erupted after the police killing of George Floyd in Minneapolis were also an important factor, in 2020, although experts differ about why. Some argue that the police, under intense scrutiny and demoralized, pulled back from some aspects of crime prevention. Others put the emphasis on the public, suggesting that diminished respect for the police prompted more people to try to take the law into their own hands.
“The distrust of police, the low morale among police, the fact that the police are being less proactive because they are legitimately worried about being backed up by their superiors” were all contributing factors, Mr. Winograd said.”
https://www.nytimes.com/2021/09/27/us/fbi-murders-2020-cities.html?referringSource=articleShare
typo
"an compelling statistical case"
should be "a compelling"
Comment deleted
Something I've always wondered about is why there isn't a massive push to hire a ton of black cops. That should help a lot right? After BLM there are extremely aggressive pushes for affirmative action in academia, the media, and so on but not in the one place where it's really needed. Electing Eric Adams mayor of NYC was a good start. But society actually has a compelling interest in the demographics of the cops matching the demographics of the population in a way that it really doesn't with the demographics of say, scientists. I'm generally not a fan of affirmative action but I would probably support black quotas in police departments at this point.
Scott, you said this "My specific claim is that the protests caused police to do less policing in predominantly black areas. This could be because of any of:
...
Police worried they would be punished so severely for any fatal mistake that they made during policing that they were less willing to take the risk" <-- I would add a related option... Police worried they would be punished so severely for any action, even if legal, they were less willing to take the risk.
Even shooting someone grabbing your gun gets your picture in the paper, a long suspension, protestors outside your house, and the mayor condemning the white supremacy of your profession. Who wants to stop a minority and frisk him, even if you can see what looks like a gun? People are going to start filming and chanting as you do it.
"Most of these points have already been made in right-wing sources, but have gone unnoticed because respectable people don’t read right-wing sources."
Do respectable people read this blog?
My own work on this:
“ This study investigated whether homicides increased after protested police-involved deaths, focusing on the period after Michael Brown’s death in Ferguson in August 2014. It also tests for effects of legal cynicism by comparing effects in homicide and aggravated assault on the assumption that reporting of the latter is discretionary and police abuses may make communities reluctant to notify police. Using FBI data from 44 U.S. cities, homicide and assault rates from 2011 to 2019 were analyzed using an interrupted time series design and combined in a meta-analysis to calculate pooled effects. A meta-regression tested effect moderators including external investigations and city/county sociodemographic characteristics. With a conservative threshold of p ≤ .01, 21 of the 44 cities experienced a significant increase and one had a significant decrease. The pooled effect was a 26.1% increase in the homicide (99% CI: 15.3% to 36.8%). Aggravated assaults increased above baseline, though the effect was 15.2 percentage points smaller (99% CI: –26.7 to –3.6) than the effect in homicides. When outcomes were measured as percent change, there were no significant effect moderators, but when measured as absolute change, homicides increased to a greater extent when the death was subject to external investigation and in cities with higher Black populations, poverty rates, and baseline homicide rates. The findings suggest that protested police-involved deaths led to an increase in homicides and other violence due to the distrust fomented within the very communities whom police are meant to protect.”
https://journals.sagepub.com/doi/abs/10.1177/08862605211028315
If policy Q leads to significantly more black people murdered compared to policy P, then my respect for the substance rather than the rhetoric of anti-racism seems to leave "more P" as the only morally viable choice to support.
Can both protests and homicide spike amoung blacks be caused by the same event? Obvious hypothesis is that the outrageous murder made lots of people very angry, some people ventilated this anger through protests others in a different way, including other murders. The distinction here is pretty important because in one case we would like to have as less protests as possible even when outrageous things happen, while in the other we would have the opposite incentive.
THANK YOU!!!
I don't think there's any better way to convince or persuade (those that can be) than to calmly, and repeatedly, 'look at the facts', "insist on thinking on a completely literal level", and walk thru careful reasoning about what's most likely true.
Have not been here in some time. Began reading the post and immediately felt a cringe. Seems all too common these days is to present various data to support various conclusions that are left hanging without context. And thus, we've generally become mistrusting of any presented data, assuming (often too quickly) a cynical attitude about where the presenter intends to lead us. Maybe that’s the point?
I'm not sure what conclusions we are intended to draw from this presentation, but I fear that any conclusion drawn has the potential to reinforce very negative stereotypes... in this case regarding race. Can this post lead to a more insightful and nuanced discussion about why murder rates rose in these areas, in these communities, that includes a comprehensive and objective accounting of this nation's history toward black communities? Might also be worthy of discussion to understand how a generalized fear pushes white folks to embrace these stereotypes so readily.
Why bring in the RK riots --which actually were riots!--when the BLM protests had a grievance and an organization behind them?
RK riots were confined to one location. The BLM movement had 7305 events in various cities across the country, the vast majority of which were peaceful.
Do the Right Thing was a movie, not RL. Spike Lee's character thinks hard for a moment before launching that trash can through Danny Aiello's store front window. We could debate what Lee is asking his audience to consider about the urban black's situation and race relations during this period when he performs this act, but that's as far as I'd go. You seem to be suggesting that art is imitating life and so we can use this film to what? Draw an analogy not to the motives behind BLM, but to justify using the word "riot"?
I'd argue that Spike Lee's character is provoking a riot out of frustration. This was the 80s. Today, there's enough statistical evidence and terms (i.e., structural racism) to create an agenda, demands, and a call to action.
Employing a word like "riot" when referring to a complex situation that has a history of grievances attached to it seems to me to be a way of dismissing that history and those grievances--we were listening until they rioted, now we know that we don't need to listen. Riot puts all of the weight on the black people attempting to address an existential problem that they now have the tools to articulate, but when you employ this word you are relegating them to all action and no thought.
Were there pockets of violence during these protests? Yeah, sometimes, but the larger movement eschewed that violence.
I'm not clear on how you are exercising "linguistic valence" here when you are using the very word that conservative news organizations picked to describe the BLM protests.
Regarding "linguistic valence," I'd say I'm operating in that field by addressing the RK riots as riots while seeing the BLM events as something more complex and thoughtful. You just keep using the same word.
Of all the sad phrases in the english language, the saddest is this:
"Steve Sailer was right (again)"
I highly recommend this blog for an 'inside baseball' view of concrete details of policing; some posts:
* https://grahamfactor.substack.com/p/how-depolicing-happens
* https://grahamfactor.substack.com/p/the-bad-cop
* https://grahamfactor.substack.com/p/its-not-about-police-reform
The author of this blog has been commenting here in this post himself, as "Graham", and on a few other recent posts too on ACX.
Hmmm, there must be more to this but rather amazingly the effectiveness of police to solve murders has been droping. It took a hit in 2020 but more amazing to me is that it's been declining since the 1960's and 70's. In particular, we saw police buildup in the 90's while at the same time tools to solve murders have gotten better and more available. A lot of stuff we take for granted today like cameras almost everywhere, people's cell phones producing logs of their geographical locations over time, DNA testing, a lot of forensics, this all should have made it a lot easier to solve murders. Yet police effectiveness here seems to be falling. Are murderers getting more cunning, super-villian types that only a Bruce Wayne or Sherlock Holmes can figure out? I doubt it.
If police effectiveness is going down, then any theory that BLM protests increases crime by causing police to pull back is flawed. If police are less effective, then it matters less and less that they pull back.
I suspect what you're seeing is a gradual decline in social trust, in particular between communities with high crime and police. The reality is probably half of murders are brain dead easy to solve, the police literally roll up on the killer covered in blood holding the murder weapon. Beyond that social trust does a huge amount of work AND things like CSI science can only substitute for that at the margins. Incidents like Floyd and the pandemic accelerated and confirmed that trust was lacking.
Less that BLM protests magically cause crime, more like they are both symptoms of the same underlying dynamic.
https://twitter.com/chrishnews/status/1542173173008957441
Whether decreased policing leads to increased crime, and if so the effect size and exact type of policing that reduces crime, is a pretty open empirical question. That said, regardless of the causal mechanism, it sure seems like the crime increase of the BLM protests outweighs the minor police reform that followed (though perhaps it sets the stage for more police reform down the road... long-term consequences are always intractable). This is not to mention the political consequences of Democrats losing votes due to Defund and thus an increased chance of our democracy collapsing, less climate change action, etc.
This was not obvious at the time though. There was a brief period where it looked like we might be getting actual police reform: Minneapolis looked like it was going to remake their police force, a bunch of red states passed (very) minor police reform legislation, etc. Support for BLM and racial justice went significantly up for a brief period, figures which had been stable for a long time.
What does this means for future BLM protests? It is clear that if you are going to have them at all, make them super policy-focused, to actually get something for our troubles. But at this point it's probably not worth having them at all in the foreseeable future, as all they do is polarize the debate around left-wing Defund.
Nice piece!
It was pretty clearly the pandemic lockdowns. It makes sense that crime didn't spike immediately but instead took a couple months before people really got cabin fever.
In support of “it’s complicated”.
The Minneapolis police killing prior to George Floyd was that of a white woman by a black cop,
Noor - the black cop - was convicted of manslaughter.
https://en.m.wikipedia.org/wiki/Killing_of_Justine_Damond
AFAICR this was the first time that a jury decided that a Mpls police killing was murder.
A couple years later, Derek Chauvin was the second.
Really, the thrust of this post seem to imply that the BLM movement is some sort of first cause. There was a long history of fear and anger about unequal treatment of non whites by Mpls police.
It became common in the Twin Cities to hear of a police shooting and respond with the question “Driving while black?”
Again I’ll say the violence that followed Floyd’s death was not justified. A lot of the people taking part didn’t give a damn about Floyd. There were many people - mostly white - taking advantage of the opportunity to indulge any malicious impulse that occurred to them. The police were in a crouch, all bets were off. What better time to plunder the OxyContin shelf at the local pharmacy? To pin all of that behavior on the BLM movement is just wrong. One of the felony conviction from the early period was against a masked white biker knucklehead smashing storefront windows with a 4 pound hammer, just out for a little recreational mayhem. Another was a white kid who was so dim that he brought police tactical gear that he looted from a police station to work to show off as souvenirs.
Hell, the only houses in my rather sedate St Paul neighborhood without a “Black Lives Matter” sign on their lawn are mine and the black couple across the street.
Murder increased because the police pulled back after the George Floyd riot? Yes, I already knew that and fully expected it.
But it’s flipping complicated.
https://en.m.wikipedia.org/wiki/List_of_killings_by_law_enforcement_officers_in_Minnesota
“Here’s what the Minnesota Department of Human Rights discovered in it’s investigating of MPD”
https://www.kare11.com/article/news/local/minneapolis-police-department-human-rights-investigation/89-82ed0eeb-ac93-4c8d-a55c-7f887eb63c9c
Cost-benefit calculations should not look at only the costs and not the benefits. If your proposed mechanism of action is that the protest caused reduced policing which created a cost in the form of a spike in murders, then I think it would've been cooler if your analysis also included the benefits of reduced policing. What are the costs of policing?
For instance, if you're using reduced arrests as your proxy variable, we should ask how many arrests are we willing to pay in order to prevent one murder? In 2019 USA there were 10,085,207 arrests and 16,669 murders. In 2020 USA there 7,632,573 arrests and 21,570 murders. The ratio of the differences come out to 500 arrests per murder.
Is that worth it? Maybe! Depends on how bad arrests are compared to murders, and what other metrics are at play. I'm not claiming to know the answer, I'm just saying that I wish that this analysis had touched on relevant questions like this. (I get that you're responding in the context of countering the media and all that, but still ... I really feel that considering costs and benefits together and not separately is a good principle to stick to.)
This is just a back of the envelope calculation of course - a real analysis might look at stuff such as prison-life-years welfare metrics, economic losses, etc. And of course, none of this is really establishing causality super well. I'm just making a more general point that even if we grant your model of this mechanism, in order to draw useful conclusions from this analysis about whether the protests are good or bad, the costs of the intervention (here, the protests) must be considered in conjunction with the benefits.
I think it's really, *really* irresponsible to express "homicides spiked after the BLM protests" as "BLM protests *caused* the homicide spike." For one thing, there is no way to demonstrate that homicidal-ness somehow increased due to BLM (I seriously doubt that it did), which is what that assertion implies. Moreover, you say yourself that a police pullback led to the increase. So what gives? You already identified the real answer: police stopped doing their jobs. How is that BLM's fault? Cops were never seriously threatened during the protests. As someone who was at them, it was very much the other way around. You've laid out a story where cops were fleeing well-deserved scrutiny, and the result was less deterrence of crime. But this isn't BLM's fault, this is the cops' fault for being bad at doing their jobs under scrutiny. Phrasing the headline as if it was the fault of innocent protesters that crimes *unrelated to the protests* went up is incredibly irresponsible and a massive show of disrespect to people like me who want to see the policing problems raised by BLM taken seriously.
I have a friend who claims that one of the reasons that the BLM thing blew up so much, at least in the sense that left-leaning institutions like academia and the media talked about it, is that the Democrats wanted to use it to increase black turnout and oust Trump. How plausible is this hypothesis and how would we test it?
Calling the BLM protests "the cause" feels like a bit of a stretch - yes, the data shows good support for "contributed". But either one plays the cause game all the way back (police brutality leads to protests etc.), or one acknowledge that the police making agentic choices is closer in the causal chain to the measured outcome.
When you have a causal chain, what about the BLM protests specifically makes it the thing to single out in the chain? My guess is that there is some empathy and less agency attributed with the stance "of course the cops reduced their policing"; but "of course they protested" has less support. Why?
Seems like police behavior in response to the protests as a plausible theory of the cause in the spike of inter-personal violence should be addressed in a lot more detail within this essay that concludes it probably was the protests themselves. That, after all, is what the Ferguson effect is supposed to refer to: People protest the police, and in response, the police refuse to do their jobs. This reduces the deterrent effect of proactive policing, which leads to more crime.
The lesson of such a theory isn't necessarily to stop protesting police misconduct or problems with criminal justice. You could instead try to break the chain of police engaging in traditional striking behavior in response.
Of all the foreign countries mentioned Honduras is the most instructive. Honduras has the highest highest documented murder rate in the world. Other places are probably more violent but they are so chaotic reliable data are hard to come by. In response Honduras banned the private ownership of any firearms. The murder rate continued to climb. This debunked 2 libtard fantasies. That gun control works & that government can make something disappear by banning it. The government tried to ban alcohol. That worked out great didn’t it. The government banned heroin well over a century ago. You can’t find that stuff anywhere can you. Ok it’s been largely replaced by fentanyl an even more dangerous drug but the point remains.
I am wondering about the use of the word "caused" (rather than, for example, "triggered") by the protests. The post unpacks a potential complex of interrelated phenomenon under the heading of the Ferguson Effect that would explain how anti-police protests set off a set of actions that result in less policing (and possibly less cooperation between communities and police) and other effects. So the pattern would be expected. But to say "caused" carries the sense of a more direct outcome rather than a distal outcome. The difference is significant if one is thinking about how to intervene in a social pattern or whether intervention is possible.
Dang, I remember when this kind of talk (libtards) was called out on this site. Like Stewart Lee, this site really has let itself go.
One of the many difficulties with crime data and making inferences on it is the chain of causal effects that occurs before data is collected. For any given effect seen in the data, you generally have to ask: is X going up because more people are doing X, or is X going up because police are paying more attention to X? Murder is actually just about the only crime we can undeniably say is happening more or happening less at any given time, simply because it comes with a body and we're generally pretty good at reporting on bodies. We can say, undeniably, there was a spike in murders after the George Floyd protests - you can literally just look at the data and know for a fact that yes, that happened.
It's everything *else* that you can be much less sure of. Sure, the spike happened *after* the George Floyd protests... but was it *because* of them? That is a much, much harder question and I don't think you've at all answered it here.
There's a few questionable assumptions and assertions being made about the data here. But ignoring all of those and taking absolutely everything said here to be true, are you sure you've drawn an exclusive causal chain? Because I think I could lift most of this evidence as written and instead make the opposing claim: the stress of seeing a black man brutally murdered, in public, by an authority figure drove an increase in homicide nationwide, predominantly centered in black communities where that stress was felt the most. We need not invoke protests or a hypothetical 'police pullback' at all.
I guess what I'm saying is: did you find the data because you had a viewpoint you wanted to prove, or did you let the data lead you to the viewpoint you're expressing here?
The fact that empirical researchers are hesitant to draw a causal conclusion on what caused the spike is specifically because that's the *right* way to interpret data that is literally inundated with idiosyncracies during a year that was itself literally inundated with novel events.
Why were murders also up in rural areas w/out BLM protests? https://www.wsj.com/articles/violent-crime-rural-america-homicides-pandemic-increase-11654864251
I don't understand why you're so focused on policing changes as causal. If you're right that the increase is predominantly black-on-black in significantly black areas, it seems like level 1 analysis should start with.. black people, not cops. And if black people were more pissed off and angry- not because of the protests, (largely) not because of BLM, but BECAUSE OF WHAT INCITED THE PROTESTS TO BEGIN WITH- then you'd expect more violence ceteris paribus.
"Cities where large numbers of protesters turned out, relative to their populations, decreased police expenditures by an average of $12 per city resident. Los Angeles’s relatively high protest rate translated into a $2.50 per-capita cut in police spending. Minneapolis had an even higher protest rate of 77 participants per 1,000 residents, which coincided with a $32.50 per-capita chop to police spending. (For reference, cities spent an average of $422 per resident on law enforcement in 2021.)"
https://www.washingtonpost.com/politics/2021/06/12/did-last-years-black-lives-matter-protests-push-cities-defund-police-yes-no/
The monetary reduction seems pretty small relative to the size of the effect being observed.
Maybe there certain large fixed costs that should be considered that would make the per-officer-on-the-street cost larger compared to the cuts.
"you people".
People keep saying things like “oh, I liked when he made an compelling statistical case showing that the media was completely wrong on this one thing they sounded very confident in, but then he started saying the media is often wrong and biased, and that sounded cliched and conspiracy-ish and right-wing, so I lost interest”.
People keep saying a lot of things that are wrong, so don't worry about it. If you sound conspiracy-ish and are correct, so be it. In other words, I think it's best to optimize for accuracy rather than for telling people things that won't put them off. We get too much of the latter already and not enough of the former.
What this doesn't really address is the surge in violent crime in rural areas:
-----
Homicide rates in rural America rose 25% in 2020, according to the Centers for Disease Control and Prevention. It was the largest rural increase since the agency began tracking such data in 1999. The CDC considers counties rural if they are located outside metropolitan areas defined by the federal government.
The rise came close to the 30% spike in homicide rates in metropolitan areas in 2020.
The CDC hasn’t analyzed 2021 homicide data yet. In some rural counties, murder rates remained high last year, while in others they have begun to recede along with Covid, data from local law-enforcement agencies shows.
...
In cities, law enforcement and civic leaders have blamed the increase in violent crime on factors such as police pulling back after racial-justice protests, the proliferation of guns, initiatives to release more criminal suspects without bail and a pandemic pause in gang-violence prevention programs.
In rural counties, where ties between police and locals are often less fraught, officials say the reasons for the rising violence are hard to pinpoint. They speculate that the breakdown of deeply rooted social connections that bind together many small communities, coupled with the stress of the pandemic, played a role.
Pastors point to the suspension of rituals such as in-person church services, town gatherings and everyday exchanges between neighbors. Such interactions can serve as guardrails, helping to prevent conflicts from turning violent. The psychological and financial stress due to isolation and job loss were especially pronounced in remote areas, where social services were limited even before Covid-19 struck, local leaders say.
As the pandemic took hold in the spring of 2020, fights between family members, acquaintances and even strangers escalated more frequently into deadly confrontations, authorities in some rural counties said.
-----
Obviously, causes could be different in different areas, but they could also be similar or the same in rural as well as urban areas.
But this seems to lend far more credibility to the idea that these trends have many different causes and that they don't fit nearly into any one box.
https://www.wsj.com/articles/violent-crime-rural-america-homicides-pandemic-increase-11654864251
Correlation is not causation.
Sorry, I know that’s a cliché, but it became a cliché for a reason, and it definitely applies here. You literally have nothing but a correlation in time (and an appeal to your personal priors) and you’re using it to dismiss people who say the factors are complex. (And those people are not media editorial boards, as your phrasing implies. They’re named experts quoted in news reports.)
If we’re going to play the game of pointing at graphs and using our common sense to make something up, I think I can do better: The BLM protests and the murder spike shared a common cause – the widespread conviction that police didn’t care about the murder of black people and even did it themselves at the slightest provocation. After all, which would do more to make you feel you could get away with murder? Learning that police don’t care or learning that they won’t be driving around in their cars for a few months?
My raw speculation attributes the spike to a combination of high unemployment and stimulus cash at the time. Lack of a job gives plus a handful of cash led many people to venture into the business of buying and selling illicit drugs. But since there is no court to resolve disputes in that business, these new players found themselves on the wrong side of a gun.
Now the next subject of research should be whether the police in Uvalde were reluctant to act because of the BLM marches in their community in the previous year -- and they did take place -- and whether this isn't the question to ask if Latino police failed to act to suppress a Latino shooter. Still, it's important to research this issue as a school shooting is the most dramatic form of murder short of war and if there were any campaigns or policy changes or funding issues that could contribute to that phenomenon, we need to know it. No good saying that there were school shootings before the pandemic or BLM. THIS school shooting in Uvalde distinctly involved police unwilling to act forcefully, and blaming this on faulty comms or garbled reports when 45 minutes go by doesn't cut it.
I'm very confused by the emphasis of this essay. The narrative it presents is: police kill black people at a disproportionate rate; people protest against that; in response, police shirk their duty and oaths to keep people safe. If that's true, isn't that a *horrible indictment* of the police? In which case, why is it "politically inconvenient" for liberal media? And why would you say that the *protests* caused the homicide spike, rather than the police refusing to do their jobs?
Should a mixture of two ingredients cause an explosion, the explosion will follow the addition of the second ingredient, but that doesn't exonerate the first ingredient from causal role.
Some logic problems:
> and in the end focuses on rising gun sales (but guns are mostly bought by white people, and so can’t explain why the homicide spike was so overwhelmingly black).
Unless more guns in the hands of black is influenced by guns passing down from white legal purchasers to blacks getting guns through illegal trade. I'm not saying that's the explanation, just that it should be accounted for in your explanation of causality.
> I think there’s clear evidence that the current murder spike was caused primarily by the 2020 BLM protests.
Again, "caused" is problematic. You, yourself suggested the "cause" is cops doing less enforcement. Perhaps less enforcement is "caused" by the protests, or maybe it's "caused" by the murder of George Floyd? Or maybe it's "caused" myriad factors that contribute to hundreds of years of discrimination that led to Floyd's death?
How, along the causal mechanism, do you point to a singular "cause?"
> (Minor quibble – our philosophy isn't "Rationalism" – it's more accurately 'modern rationality' (no capitals) and it's less of a _compact_ 'philosophy' than a much more nebulous 'cloud' of ideas/aspects from all kinds of prior/current bodies of knowledge.)
- I appreciate you nuancing, though I can't say I understand the definition you are trying to express
> "I don't like that first part – you will _definitely_ "always find things to be critical of Scott"? Even if Scott states "I agree with Jordan."? That just seems _deliberately_ contrarian :)"
- I don't like it either because that's not what I intended to express lol. I meant it in the sense that people are different people, I know of no two people fully agreeing on every topic. I was ascertaining my right to disagreement, not my right to being a deliberate contrarian. As an aside, I do like stepping in other shoes for the sake of debate and taking on a contrarian position. The farther from my own beliefs, the more entertaining and thought provoking I find it.
1. I think not enough consideration has been put into the details of the mechanism by which police pullback supposedly caused more murders. It's possible, but how - exactly - did police work stoppages lead to more murders? I think that aspect of the hypothesis matters.
2. There's very clearly an uptick in Native American murders. Handwaving that away as statistical noise undermines the credibility of your argument, I think - it comes across as sweeping data that contradicts your hypothesis under the rug.
So, your analysis is that…. cops stop doing their jobs when publicly criticized? And *specifically* abandon their duty to serve and protect Black communities because they’re mad about and afraid of being critiqued?
I was inspired reading Scott Alexander's piece "What Caused The 2020 Homicide Spike?" So much so that I included it with 7 other primary resources proving there was a Ferguson Effect, where attacks on police and protest led to further bloodshed.
It's a long piece written the past 24 hours, but I excerpted the section where I focus on the Astral Codex Ten article from a couple weeks ago. Let me know what you think.
https://agent-orange-chicago.medium.com/evidence-of-a-ferguson-effect-costing-thousands-of-ameican-lives-349f606faf94
Is It The Pandemic?
If you think murders increased from the “pandemic,” another hypothesis thrown in here, then you’re also off. Murders fell or remained static across Europe, Central America, and South America, and I’ve yet to find a single country that had a 2020 murder spike except the country (USA! USA!) that had thousands of “mostly peaceful” anti-police protests after George Floyd’s death, 7% of which turned violent. (Again, a situation where all the Minneapolis police involved were immediately arrested. So much for “no justice, no peace.”)
But critics say “more research is needed.” Which I’m all for, but somehow they missed the multitude of studies indicating a “Ferguson Effect” can be shown statistically and with strong methodologies. I don’t think that’s an accident that they ignore brave academics like Wilfred Reilly, Justin Nix, and Roland Fryer.
Is It Weeks of Months of Anti-Police Protest, Civic Disruption, and Attacks on Policing? (Bingo.)
Here’s a vastly more intelligent and thoughtful analysis by famous rationalist and blogger Scott Alexander. He also had a brouhaha with The New York Times, where he has legitimate beefs with them trying to out him publicly and tarnish his reputation, but I won’t get into it.
Simply, the creator of Astral Codex Ten is not beholden to media or academic pressures, where merely suggesting the “Ferguson Effect” means you’re not an “ally” and possibly “racist.” Alexander puts the near 5,000 single year increase in murders squarely on the BLM protests. Is there any incentive for career-minded academics or media journalists to promote this theory, even if painfully obvious? Of course not.
Read for yourself:
https://astralcodexten.substack.com/p/what-caused-the-2020-homicide-spike
Excerpt:
Conclusion
I think there’s clear evidence that the current murder spike was caused primarily by the 2020 BLM protests. The timing matches the protests well, and the pandemic poorly. The spike is concentrated in black communities and not in any of the other communities affected by the pandemic. It matches homicide spikes corresponding to other anti-police protests, most notably in the cities where those protests happened but to a lesser degree around the country. And the spike seems limited to the US, while other countries had basically stable murder rates over the same period.
I understand this is the opposite of what lots of other people are saying, but I think they are wrong.
Scott Alexander is brave, but he’s not alone. If more people stand up against the ideological grain, I honestly believe lives can be saved.
xx end excerpt xx READ HERE:
https://agent-orange-chicago.medium.com/evidence-of-a-ferguson-effect-costing-thousands-of-ameican-lives-349f606faf94
The timing made it pretty clear in most cases. I was looking through the data for Oakland, CA, and the spike was very large and happened almost overnight the week after the Floyd death. No other explanation could lead to such a quick shift like that. Pandemic, gun sales, poverty, change in the weather.....all other explanations would surely have more gradual increases and at different points in the year.