Do you have a dumb question that you're kind of embarrassed to ask in the main thread? Is there something you're just not sure about?
This is your opportunity to ask questions. No question too simple or too silly.
Culture war topics are accepted, and proposals for a better intro post are appreciated.
Jump in the discussion.
No email address required.
Notes -
https://substack.com/app-link/post?publication_id=89120&post_id=187622533
Anyone see Scott's latest post on crime stats and have a significant disagreement with it? It feels culture war worthy and I'd like to discuss it, but I don't feel like I know the counter viewpoints well.
Some criticisms:
People in cities don’t report their car being broken into, and this rate of reporting will decrease as social trust decreases. So the property crime metric may not be accurate.
The National Crime Victimization Survey is a mail-in survey. People living in low-trust (high-crime) areas of the country increasingly have no interest in filling out government surveys, having no trust in the government and having less familiarity with filling out mail. It would not be “an extraordinary coincidence if they exactly matched the proposed reporting bias to police”, as Scott asserts, because habits follow environment — those raised in a high-trust environment will fill this out at a higher level, and high-crime areas gradually reduce in trust.
One reason for the decline in car thefts may be that, on average, they have been more difficult to steal and get away with the theft due to omnipresent surveillance.
It’s strange to me that people trust surveys from phone calls, emails, and the mail. Academics must have zero familiarity with how the lower class and lower middle class feels. It would be better to go to Walmart and run a survey, or hop on voice chat in a popular video game and ask the users if they would ever in a million years fill out a mail survey. They really think someone smoking weed every day in a high crime area is going to fill out a government survey or answer a government questionnaire (in their mind no distinction between an institute and the government)? These are not accurate assessments of a bulk of the population, and they are increasingly inaccurate. You are surveying the most productive remnants of the common flock who still possess the time, trust, sense of social duty, attention, and conscientiousness to answer your survey.
This would be a good study: go to a lower-income school and and make them answer some questions from someone who looks to be in the same class. Ask them if they would ever fill out out a government surcey or answer a phone number they don’t know or answer an email survey. Some second-gen Asian students who happen to live in the area will say yes and everyone else will say no.
With that said, I do think crime is decreasing, and that the crime rate is a red herring wrt what people about (“is there a menacing person playing his music loudly on my commute”)
I wonder if we have data on the average response rate for such surveys? Are they getting as many responses back in 2025 as in 1990? If that’s the case then my hypothesis is disproven, but if it’s the case then they have to narrow it by income and test hypothesis
More options
Context Copy link
I think there's an inherent weakness to these population-wide metrics. Let's say there's a gang-infested Gotham city with 1M people and there's a suburb next to it called Boring Creek where another 1M people live. And let's say there is 1000 murders per year in Gotham and 1 murder per year in Boring Creek. Statistically, we have 1001/2M = 0.5 per thousand murder rate. Now, let's assume due to Batman's effort in Gotham, the murder rate there decreased to 500. While Boring Creek voted to defund the police and convert it to "social crisis nonviolent intervention consultants" and now their murder rate is 10 per year. Now we have 510/2M = 0.25 per thousand murder rate, or half as bad as before. Technically true, but Gotham is still a hellscape with 500 murders per year, and Boring Creek's resident chance of being murdered is now an order of magnitude larger than before. So, if the radio host of Boring Creek Patriotic Voice says the town is going to hell, and the host of More Facts Than You Would Ever Know substack says we're winning the war on crime like never before and the Patriotic Voice are just innumerate idiots who don't care about the facts, who among them is right?
I mean, it's good that murder rate over the whole nation goes down. But how does it go down is important too. If it goes down in a way that out of 5000 shootings in Gotham previously only 4000 survived, and now due to advances in emergency medicine 4500 survive, does that really help me that much, if I live in Boring Creek? And even if the gang on Gotham are living in mortal fear of Batman and actually are shooting each other less - does that really help me that much, if I live in Boring Creek?
More options
Context Copy link
I think there's room to ask about whether, even as the crime rate-per-population has gone down dramatically, the rate-per-potential-exposure has been less changed or has gone up. As Scott says, "We’re a safetyist culture"; we avoid risks more than we used to. We also have more attractive alternatives to risks - where I would play sports in the street or at worst play video games in person with the neighborhood kids, my children go to the rock-climbing gym or play networked video games with their friends farther away. I grew up in a residential area where once I got old enough I could walk to a convenience store, perhaps past some sketchy houses; my kids are growing up in a giant suburb where it wouldn't matter if the houses were sketchy because there's nothing they could get to on foot regardless.
On the other hand, the answer might just be "no, the rate-per-potential-exposure has gone down too". Or it might be that this isn't a sufficiently well-defined metric, because in a big country there's always someplace where it's just too dangerous for an innocent person to go and someplace else where it's perfectly safe and there's no obvious way to decide how to weight those places when averaging.
If the murder rate stays constant, but “rate per potential exposure” gets worse,
someone is getting exposed at a higher rate.the people who are getting exposed must be making up the difference. Who? Shouldn’t it be strictly easier to tell which neighborhoods have turned into death traps?Just the opposite.
murders / population = murders / exposures * exposures / population. Ifmurder / populationis constant whilemurder / exposuresincreases thenexposures / population, the exposure rate, must be decreasing inversely.Is it? I know there are sites that give neighborhoods "walkability" scores, but at least the first one I pulled up is only giving a theoretical number based on the mass transit availability, distances to the nearest grocer/cafe/school, etc; I'd have no idea how to find an actual number of people who walk down a particular street (or who drive in a particular area - the only armed robberies I found out about first hand were at a stoplight and in a parking lot) on an average day.
Whoops. I was trying to gesture at the conditional probability P(M|E). If P(M&E) was constant/dropped, but P(M|E) went up, then yeah, P(E) must have dropped.
I don’t know why it dropped. Maybe the walkability scores really worked, and following them is enough to dodge almost all crime. Seems unlikely. Maybe law enforcement drove most criminals into hives of scum and villainy, and now word of mouth is enough to keep tourists from visiting Skid Row. Maybe COVID killed all the criminals first. Any number of stupid reasons.
But people aren’t acting like P(E) has improved, are they?
I am tempted to argue that this is a media phenomenon. That if people weren’t getting pictures of immigrants piped to their phones 24/7, they wouldn’t feel like P(E) was so high. I’m aware that this flatters my own biases, so I’ll try to discount it, but surely something like this is possible.
As someone else mentioned, I think perception of crime rate is highly influenced by visibility of antisocial behavior. Murders and car thefts have fallen, but if going in public involves navigating multiple individuals screaming at invisible demons and flailing around, the drop in murders is probably not that comforting. The media undoubtedly plays a role since they're the ones telling me about the knife attacks (but non-fatal!) on public transportation in my nearby city, but I can see the homeless encampments and sketchy methheads with my own eyes.
Exactly, I feel more impacted by crime in comparison to twenty years ago, even if the really bad stuff was quite a bit more likely back in the day.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link