What the data says about crime in the U.S. | Hacker News Reader

What the data says about crime in the U.S.

pewresearch.org

49 pointsgmays80 comments

The article provides an alright overview but omits a critical aspect: the correlation between race and crime rates is strikingly high. For example, FBI data shows vast differences in murder rates by race:

African American: 33.6 homicides per 100,000 people American Indian and Alaska Native: 12.9 per 100,000 Hispanic: 6.9 per 100,000 White: 3.3 per 100,000 Asian and Pacific Islander: 1.7 per 100,000

There are other crimes in which the numbers are flipped, such as gambling and financial crimes.

If we're serious about making things better and stopping violence, or whatever crime, we need to be upfront about who it's impacting the most and why.

I think a lot of crime Statistics are twisted and misreported to meet political agendas. One side doesn't want to give the other side any ammunition.

This is the case with literally everything and every large scale social argument, both sides selectively choose facts that align with their agenda. This is why newspapers which have editors (that selectively choose stories) and writers (that selectively choose what facts go into those stories and what "context" to provide) are propagandists and always will be.

Now that we have techonlogists who are replacing both with algorithms that pattern will extend most likely, to the algorithm designers.

Now correct for socio-economic status and watch those racial disparities almost disappear.

You sure about that almost part? I don’t really know but I want to make a different point.

Stop collecting race and ethnicity data! It’s really that easy. France long ago outlawed the very collection of such information.

Why? because use socioeconomic analysis and variables that exclude race to have compassionate policies that treat all humans equally. Once you inject race and ethnicity into policies it does nothing but harm, it either provides excuses or fodder for racists, just outlaw the collecting of that information, France has for a long long time. Collecting anything else, any other socioeconomic variable is fine and can be used to create policy without such negative consequences.

Its almost as if the policies produced by racist conclusions create the conditions that perpetuate the misleading data. /s

Humans are an "INBRED" species. By this I mean we have little genetic diversity compared to most other species on the planet. Race isn't even real genetically speaking. It is real in a socioeconomic sense because people believe its real and that has real consequences. But outside of human belief and behavior, race is a load of imaginary divisions.

I have zero evidence to support this, but I would be more inclined to believe that (sub) cultural environments impact these stats way more than genetics. None of the data or even logical reasoning supports a genetic component.

> Race isn't even real genetically speaking

If you map every human genome in some space, there will be clusters. This is race.

You can argue that there are no clear boundaries between clusters but that's true for everything. You can use that argument to say there's no difference between trees and bushes (and so on, applied to every distinction).

Genetic clustering segregates people into a different set of groups than the ones that are involved in discussions such as this one. It's a different definition of "race".

> If you map every human genome in some space, there will be clusters. This is race.

There isn't a reason to think that those clusters would in any way map on to how we culturally, currently talk about race. Nor is there a single, well-defined way to cluster human genomes plotted in some very high dimensional space.

> You can use that argument to say there's no difference between trees and bushes

That's actually a perfect example. We have a word in English—"tree"—that refers to a large number of different species of plant. And you could plot plant genomes, and do a cluster analysis on them. But there will be no cluster that corresponds to the english word "tree". Genetically, there isn't a single "tree" ancestor, that all trees descended from. Instead, what we call "tree" evolved repeatedly at different points in time, and so no matter how you cluster the genes, what we refer to as "tree" will always be spread out among different clusters.

> There isn't a reason to think that those clusters would in any way map on to how we culturally, currently talk about race.

Of course there is. However loosely, "cultural race" is linked to "genetic race".

Regardless, the post I replied to said "Race isn't even real genetically speaking".

> Nor is there a single, well-defined way to cluster human genomes plotted in some very high dimensional space.

There's very clearly an simple way to talk about genetic differences between groups. I don't know if there's a "single, well-defined way" but there doesn't have to be.

> That's actually a perfect example. We have a word in English—"tree"—that refers to a large number of different species of plant. And you could plot plant genomes, and do a cluster analysis on them. But there will be no cluster that corresponds to the english word "tree". Genetically, there isn't a single "tree" ancestor, that all trees descended from. Instead, what we call "tree" evolved repeatedly at different points in time, and so no matter how you cluster the genes, what we refer to as "tree" will always be spread out among different clusters.

The point I was making with the trees/bushes aside had nothing to do with whether we define trees genetically. It had to do with the fact that "fuzzy boundaries" or "lack of a single, well-defined boundary" isn't a good reason to discard a distinction.

Moreover, humans only evolved once.

It makes intuitive sense, but intuition is often misleading and deserves to be double checked.

In this particular case, while there does exist a set of alleles against which you can cluster populations in a way that maps roughly to the set of races we define socially, that only represents < 10% of the genetic variation across populations. Basically, that means you need to arbitrarily toss out > 80% of the dimensionality of your data for that clustering to meaningfully map against what we socially construe as "race".

This casts a large shadow on the idea that the genetic notion of "race" can serve as support for using the social notion of "race" as a means to treat anyone differently. You may be correct in the strictest technical sense, but the link is so tenuous that it can't really be used for anything useful.

"Race is a social construct" is a bit of a misnomer. The proper formulation should probably be something like "Race is only meaningful because we socially decided to make it so."

> In this particular case, while there does exist a set of alleles against which you can cluster populations in a way that maps roughly to the set of races we define socially, that only represents < 10% of the genetic variation across populations. Basically, that means you need to arbitrarily toss out > 80% of the dimensionality of your data for that clustering to meaningfully map against what we socially construe as "race".

This doesn't make sense to me. Can you provide more information?

I think you're referring to the fact that genetic differences between individuals within a group are larger than differences between groups. But that does not invalidate the differences between groups, or mean that we have to make "arbitrary" decisions to measure those differences.

I don't agree with the rest of your post but don't see the point of going into detail.

> I think you're referring to the fact that genetic differences between individuals within a group are larger than differences between groups

But that's the rub though: How do you define that there are groups in the first place? There is no way to land at our socially-defined set of races via objective means.

How do we differentiate between yellow and orange? The spectrum of colors is a continuum. Is the difference between yellow and orange not real?

> Is the difference between yellow and orange not real?

"Yellow" and "Orange" are words, not facts about the universe. They correspond (loosely) to physical observations, at least when it comes to perceived colors (there are plenty of optical illusions which reveal the vast gulf between 'the color a thing is' and 'the color we perceive'). So I don't really know what you mean by "real" here, but is it an objective fact? No.

> "Yellow" and "Orange" are words, not facts about the universe.

That's a philosophical position called nominalism. You can be a nominalist but other people disagree.

Right, the difference between what we call "yellow" and "orange" corresponds to objective physical properties. I could reject the way we have decided to group these differences, but the existence of different wavelengths seems to be a fact about the universe.

The different wavelengths is a fact about the universe, but there is nothing about the wavelengths themselves that makes one “yellow” and one “orange”. That’s all added by us culturally.

I think you're begging the question here. If you look at the differences between DNA, there will be clusters (because some people are more closely related than other people). These clusters are the groups I'm talking about.

> Race isn't even real genetically speaking

I was expecting a flood of these types of comments but also "facts are racist" and "you can't be racist against white people" and I'm not disappointed. This and seeing the post flagged just speaks volumes about where HN is heading.

I don't mind them omitting this particular set of numbers because the very strong correlation between race and socio-economic status tends to change the picture considerably when you control for it.

The raw numbers are misleading in a very politically charged way.

In what ways should society use race data to make things better and stop violence? Also wondering about the cause of the difference.

I want to make a general comment about this. I don't direct this to you, personally. I bring it up because it's a common talking point in certain pllitical leanings: the idea that race is a factor in crime (specifically, that black Americans commit more crime). It's an opportunity for media and staistics literacy. I'm glad that most of the sister comments here recognize the issue.

Let's say there is a racial disparity in crime. Assuming it's true, the reasons are either intrinsic or extrinsic. Intrinsic reasons (eg a certain group being fundamentally more violent) are racist arguments. Extrinsic reasons mean there is some other factor not accounted for in the day.

This brings us to the confounding variable [1]. There is some other factor that creates this disparity. And that factor, which other commenters have correctly identified, is poverty (ie socioeconomic status).

The link between poverty and crime has been known for, quite literally, thousands of years. Aristotle is known for saying "Poverty is the parent of revolution and crime". This fact is often overlooked in modern political discourse.

I bring this up in case anyone encounters this argumment in the wild.

[1]: https://www.scribbr.com/methodology/confounding-variables/

> it's a common talking point in certain [political] leanings: the idea that race is a factor in crime

So . . . people with "other" political leanings are known to attribute many problems to poverty. Many more problems than other politics would allow.

And they call us radical leftists. We are not widely accepted. Just sayin'.

how exactly would this knowledge affect policy? all I see are people online saying "despite only being 13 percent of the population..."

We could use it to more heavily police certain neighborhoods and the police could profile people by race . . . oh! Never mind.

> There were no major differences in violent crime victimization rates between male and female respondents or between those who identified as White, Black or Hispanic.

This statement is profoundly misleading, calling into question the validity of the entire article. Black Americans are much more likely to be homicide victims than any other group. The fact that the general category of "violent crime victimization" appears to be equal for all 3 groups probably just means that there are huge differences in what gets reported. A street fight in an affluent white neighborhood probably means that the cops get involved, the same level of altercation in a majority black area probably does not involve police.

They come armed with statistical data. You come with a statement of fact that you don't support with any data. While I might be predisposed to believe you, you do not in fact offer any support for doing so.

That came from self-reported survey data, a well-known source for manufacturing statistics.

Can you share data to back that statement up please?

I think what they were getting at is that crime is more closely linked to age and economic status than race

> younger people and those with lower incomes were far more likely to report being the victim of a violent crime than older and higher-income people

Not the OP, but this Wikipedia article seems like it does a decent job of covering the subject: https://en.wikipedia.org/wiki/Race_and_crime_in_the_United_S...

Which notes a similar point:

> Researchers note that there are socioeconomic factors underlying these crime statistics, and that crime is often higher in low-income neighborhoods

I'm not sure that's true. Now, the white person may not end up being charged, or their punishment may be greatly reduced (due to race, class status, and access to council), but what makes you think that the police are less likely to get involved at the same level of violence?

I think it's possible there could be a bias against certain racial groups. An assumption that this group of people commits or does not commit crimes, higher or lower than other groups, and so that primes the interaction.

Since multiple people asked for stats on this, I'm just putting it as a response to the parent instead of responding to any given individual. Here [1] they are. It's from 2018 because it seems like at some point the FBI chose to replace nice simple tables with some really horrible UX 'data explorer' that leaves me absolutely unable to find more recent data on the exact same stat. Anyhow, black Americans made up something like 40% (eyeballing) of total homicide victims, yet are only about 12% of the population. It's extremely disproportionate.

[1] - https://ucr.fbi.gov/crime-in-the-u.s/2018/crime-in-the-u.s.-...

Ah, but there is a known strong correlation between poverty and crime, and strong correlation between race and poverty. Any conflation of correlation with causation between race and crime is exclusively an attempt to obfuscate a racist political stance.

Your comment has nothing to do with what's being discussed. The self reported survey claims that black individuals have no higher chance of victimization than other groups. One of the extremely few objective measurements of victimization data we have shows they are dramatically more likely to suffer at least some sorts of violent crime. This suggests that the survey data is likely flawed.

>> there is a known strong correlation between poverty and crime

> Your comment has nothing to do with what's being discussed

What? The paper itself notes that it does

> younger people and those with lower incomes were far more likely to report being the victim of a violent crime than older and higher-income people

The data explorer is backed by wide open APIs. The numbers are just a curl away:

curl 'https://cde.ucr.cjis.gov/LATEST/estimate/state/TX/violent-cr...' -H 'accept: application/json, text/plain, /'

  {"results":
    {"Texas Violent Crime":{"2012":408.6,"2013":410.3,"2014":406.7,"2015":412.7,"2016":433.8,"2017":437.8,"2018":412.9,"2019":421.8,"2020":446.5,"2021":453,"2022":431.9},
     "United States  Violent Crime":{"2012":387.8,"2013":369.1,"2014":361.6,"2015":373.7,"2016":397.5,"2017":394.9,"2018":383.4,"2019":380.8,"2020":398.5,"2021":387,"2022":380.7}}}

Docs are here: https://cde.ucr.cjis.gov/LATEST/webapp/#/pages/docApi

The fact that the general category of "violent crime victimization" appears to be equal for all 3 groups probably just means that there are huge differences in what gets reported.

The article is very clear that they're focusing on what gets reported, and not what actually happened.

At least in my area homelessness is pretty tied to perception of crime, if not actual crime. And it's seemingly been going up a rather solid amount over the last five or ten years.

Consider these three findings together:

1. Americans are more worried about crime;

2. Crime is down regardless of source; and

3. Concern about crime is national not local.

This is manufactured consent in action. Politicians hype up crime panic. The media amplifies that message. Why you see a disparity between local and national crime is because people see crime isn't out of control in their local communities so they think the crime wave must be elsewhere.

Perceptions of crime are sharply up both locally and nationally, but with local perceptions maintaining a higher value than local. This is perfectly expected because crime is extremely unevenly distributed.

> 2. Crime is down regardless of source;

You left out "Since 1990's" ...

Data is only reported through 2022.

Concern about rampant retail theft forcing major store closures has risen over the past two years.

Also curious whether this includes crimes in areas that have a dollar limit on what will be prosecuted. Are those even reported?

It's not an illusion that locks are appearing in retail in certain areas. That's a new thing.

I don't believe those stores would be doing that if they had other choices.

> Concern about rampant retail theft forcing major store closures has risen over the past two years.

That's what they told investors to explain losses, at least. https://www.cnbc.com/2023/01/05/walgreens-may-have-overstate...

"A top Walgreens executive on Thursday acknowledged the company may have overblown concerns about thefts in their stores after shrinkage stabilized over the last year... 'Maybe we cried too much last year,' Kehoe said. 'We’re stabilized,' he added, saying the company is 'quite happy with where we are.'"

https://www.nytimes.com/2023/01/06/business/walgreens-shopli...

"The San Francisco Police Department’s data on shoplifting did not support Walgreen’s explanation for the store closings, according to an October 2021 analysis by The San Francisco Chronicle. The analysis said that while not all shoplifting incidents were reported to the police, one of the stores that closed had only seven reported shoplifting incidents in 2021 and a total of 23 since 2018."

Is your point that you don't believe that people's concern about rising crime is valid?

Absolutely. Humans are horrible at assessing such things. Statistically we live in one of the most safe and stable societies in human history.

Especially when significant amounts of propaganda - from police unions, bleeds-it-leads media, tough-on-crime politicians, companies trying to explain their quarterly loss as anything other than "our CEO fucked up" - are in play.

Two weeks ago my daughter's bike was stolen. From her school bike rack. In broad daylight. On camera. It was a local mother and daughter that stole it.

And zero police response.

I should blame the media?

Seems like being concerned about crime is pretty reasonable when people are experiencing crime.

And a lot of people are.

I hope that you are fortunate enough to live in an area where you don't experience it. Your points would make a lot more sense to me if that's the case.

I’ve had a package stolen off my porch.

I draw precisely zero conclusions over the local, state, or national crime rates from that anecdote.

One factor that I think the article glosses over is the geographic distribution of crime. It briefly mentions some variations by state, but does not talk at all about variation by local area. It briefly mentions that the FBI says population density and economic conditions can affect crime rates, but gives no data.

> In 2022, those who are male, younger people and those who are Black accounted for considerably larger shares of perceived offenders in violent incidents than their respective shares of the U.S. population. Men, for instance, accounted for 79% of perceived offenders in violent incidents, compared with 49% of the nation’s 12-and-older population that year. Black Americans accounted for 25% of perceived offenders in violent incidents, about twice their share of the 12-and-older population (12%).

Extremely misleading paragraph. It implies that Black Americans are maligned for being criminals more than their true share of the population; but the arrest data shows that blacks are 27% of all those arrested, and more than 50% of those arrested for murder. https://ucr.fbi.gov/crime-in-the-u.s/2018/crime-in-the-u.s.-...

And you think arrest rates are unaffected by racism?

Traffic stops lose their racial disparity when the sun goes down (and cleverly, the control here was daylight savings time - the disparity jumps suddenly by an hour when DST goes into/out of effect) and cops can't see the occupants of cars as well. https://news.stanford.edu/2020/05/05/veil-darkness-reduces-r...

No I don't, but that's the data I found, and it's a lot better than pretending that all races commit crime in equal proportion. Feel free to counter with better data, like murder convictions by race or similar, instead of just pearl clutching.

Setting aside the fact that we know it's bad data, equating arrests with committing crime is another mistake.

The problem with such data is always the reporting threshold. Crime level might be high but crime doesn't get reported.

There will always be unreported crimes. What matters more is changes to crime and reporting rates. Given all data sets show that crimes have dropped significantly

- If the reporting rate for crimes has been steady or increased, then we can conclude crime has indeed decreased

- Only if the reporting rate for crimes has decreased do we not know the effect. It would be dependent on the magnitudes of each

They get into some of this data in the piece

Like I told - there is problem with threshold changes. If the thresholds are manipulated then it will lower also reported crime rates but the apparent crime rate based on personal experiences circulating in the population will paint a different picture. For example if somebody gets mugged but the value of the loss is below the threshold then it will not get reported but the crime still happened and the fact is shared in the population and this affects the perceived crime rate within the population.

The problem also: major cities have simply stopped reporting crime at all. Further, crimes go unreported because police have stopped responding to many of them.

And then you have the other neat new thing of all races being listed as white.

> The problem also: major cities have simply stopped reporting crime at all.

I promise you, no major city has ceased to report and respond to homicides, where we continue to statistically enjoy one of the safest couple of decades in American history.

And what has it to do with wide spread blatant shoplifting affecting many US cities?

You mean the stat that is...

* Frequently inflated with claims that there's a whole bunch of them suddenly unreported?

* Likely blown out to some extent by statistical outliers? https://counciloncj.org/is-shoplifting-up-or-down/

* Overstated by companies to justify store closings? https://www.nytimes.com/2023/01/06/business/walgreens-shopli...

It's exceedingly difficult to determine the actual rate of shoplifting on a national scale.

The claim was "major cities have simply stopped reporting crime at all" which is just... laughable. As is the "everything's reported as white" follow-up.

And that was the main point of my comment - we really don't know what is the actual situation.

I thought it was "wide spread" and "affecting many US cities"?

Yes, this is one particular crime where crime level might (see my original comment) be higher than the official picture presents. Even when it is not reported then people still perceive it as crime.

I promise you that major cities have stopped reporting crime to the FBI. At least one major city hid lack of crime response under administrative labels.

The FBI data is based on voluntary reporting by individual law enforcement agencies, and last year, half of US police departments, including those in big cities such as Los Angeles and New York, failed to submit data for 2021.

https://www.theguardian.com/us-news/2023/oct/22/us-crime-sta...

Seven months after the May 2023, assault, Houston police classified the incident as “suspended – lack of personnel,” an internal label placed on tens of thousands of incidents over the past decade. The classification perplexed Hill, who felt the matter had been long settled.

https://houstonlanding.org/did-hpd-neglect-suspended-cases-w...

"to the FBI" is a much narrower claim, and assaults aren't homicides. It is not suprising to me that some assaults would not be followed-up on; that's why homicide rates are much more interesting.

Numbers for the missing NYC, for example, are hardly difficult to dig up. https://abc7ny.com/nypd-crime-shootings-murders/14259597/

> The NYPD released some statistics about safety in New York City during 2023. The number of murders and shootings in the city both decreased last year and are continuing to track downward following the pandemic. New York City had 386 homicides last year, according to new NYPD data, an 11.9% decrease from 2022.

> "to the FBI" is a much narrower claim

The context is Pew research using stats reported to the FBI and surveys by the DoJ.

https://www.pewresearch.org/short-reads/2024/04/24/what-the-...

There is also the effect of decriminalization and depolicing which I did not see well addressed in the article. There are parts of the US where local government has the social leaning or at least their key donors have the social leaning to take a permissive attitude towards activity that was previously criminal. Effectively this makes activity like buying and selling drugs and shoplifting legal and, thereby, meaningless to report.

If anything the article may more likely be an indicator of the delta between crime as-is and as-reported or a cautionary tale of decision-driven-data

The types of crime covered in the article are not these types of crimes

> Using the FBI data, the violent crime rate fell 49% between 1993 and 2022, with large decreases in the rates of robbery (-74%), aggravated assault (-39%) and murder/nonnegligent manslaughter (-34%). It’s not possible to calculate the change in the rape rate during this period because the FBI revised its definition of the offense in 2013.

> The FBI data also shows a 59% reduction in the U.S. property crime rate between 1993 and 2022, with big declines in the rates of burglary (-75%), larceny/theft (-54%) and motor vehicle theft (-53%).

No, that is literally what one of the two major studies mentioned, the National Crime Victimization Survey, gets around with its methodology.

They survey a quarter million Americans every year, and don't rely on what gets reported to the police/insurance companies/etc.

Did you not read the article?

Reply on news.ycombinator.com