Hacker News Reader

Comment by Shrezzing | Hacker News Reader

The report uses data from the Uniform Crime Report. UCR records crimes reported by the police, which isn't the same as crimes occured.

It's likely the link here is between Government Budgets (including budgets for police resources), and Interest Rate. Low IR generally happens in response to recessions. Budget cuts (including police budget cuts) also happen in response to recessions. Police activity reduces with reduced police budgets. Crimes reported to UCR reduces with reduced police activity.

View on HN

However, inflation versus crime is studied across nations with academic rigor. Bias from a single source does not undermine the conclusion that inflation is a higher factor in certain crimes than, say, unemployment.

https://pinkerton.com/our-insights/blog/inflation-and-crime

"Academic rigor," especially in these social science studies, has usually come to mean "fraud" or at best "bad methodology".

Search "Reproducibility Crisis", "Dan Ariely" or "scientific fraud" and you should get the jist of why you shouldn't trust many headlines for new discoveries right now. Especially "surprising" results that go against expectations

https://www.nature.com/articles/d41586-023-03486-5

https://www.newyorker.com/magazine/2023/10/09/they-studied-d...

This is not my field, but I remember that article as completely undermining your claim that “academic rigor” “usually” implies “fraud”.

My take on the replication problem (I’m not a social scientist) is that we now think the original experiments missed a confounding factor out of ignorance or lack of statistical sophistication.

This is not how Ariely and Gino were caught, though. I wouldn’t call their research surprising at all (again, not my field), but their randomization failed. Replication should have found a different level of support and introduced skepticism of the rigor of their methodology.

>usually come to mean "fraud" or at best "bad methodology".

It's what I originally wrote. Good job cherry picking. You sure this isn't your field?

Ok, so my take on the Nature article that it painted a rosy picture of today’s dispersed research teams mutually verifying each others results. Is your argument that the bad-old days still taint modern methods?

'headlines' are not the 'result's.

You should always take media 'headlines' about science with a grain of salt.

Not just for social sciences, but for dark matter, string theory, everything.

The replication crisis is bit overblown because the 'right' want to use that to drive their own more 'biblical' view of the world. By sowing doubt about all science, like done with cigarettes.

Social sciences do replicate at rate over 50%. That is long way from nothing replicating. Sorry 50% isn't good enough, it isn't like string theory is replicating, and nobody is getting out pitch forks.

another take https://www.vox.com/future-perfect/21504366/science-replicat...

It should be noted the ‘left’ also wants to use this in order to drive their own worldviews. They are both sides of the same coin after all.

False equivalence. The Left has an agenda, sure. The Right believe they are gods chosen and are fighting a holy war.

The Right believe they are being directed by GOD. Their soul will be tortured for all eternity if they don't fight a war on earth following the will of God.

How do you argue a factual point with someone that is basing their decision on being directed by the god of the entire universe.

Basically the modern Christian Right, is fighting their own Jihad. We just don't have a good word like that to describe it.

So, I would disagree they are two sides of the same coin.

The left is maybe skewing some statistics, while the Right is ready to kill people for their god (checkout how many mass shootings are based on Right Talking points).

For clarity, the Rosenfeld study referenced in the Pinkerton article (while talking about inflation rather than Interest Rates) also uses the UCR dataset as its primary datasource, while its european equivalent, also collected from police reporting, is used for its non-American datasource.

Yeah, the section intriguing to me was:

‘ The work of the historian David Hackett Fischer identified four major instances of inflation in Western history, in the 14th, 16th, 18th, and 20th centuries. In each of these periods, violent and property crime rates increased, and then fell once prices stabilized. When inflation decreased in the early 1990s, both Europe and the United States saw a corresponding decline in crime.’

Other sources use something called the Modified Wald Causality test, which should only be convincing to someone already acquainted with the field. The modern European data comes from Interpol, which might suffer the same collection bias as USA. We’re looking at first-order changes, anyway, so I wager this causality test can be sensible, even if the incoming data suffers from a linear bias.

hm... there's some correlation definitely. most likely there's simply the common factor behind them, which is the economy. (bad times get the knives, good times make you nice)

https://i.imgur.com/vTUJMA3.jpg

>UCR records crimes reported by the police, which isn't the same as crimes occured.

No, but it's the best available proxy.

> UCR records crimes reported by the police, which isn't the same as crimes occurred.

What data would be better?

Crime survey data often removes many of the noise in recorded crimes data. Basically surveying a cross section of people and asking them if they've been a victim of crime in the last X months, and what type of crime.

> Crime survey data ... surveying a cross section of people and asking them if they've been a victim of crime in the last X months, and what type of crime.

That's helpful.

Below are FBI discussions about differences and uses for National Crime Victimization Survey & Uniform Crime Reporting

https://ucr.fbi.gov/crime-in-the-u.s/2019/crime-in-the-u.s.-...

https://bjs.ojp.gov/sites/g/files/xyckuh236/files/media/docu...

A survey doesn't sound very accurate, since some people are more and less likely to be victims of crime. If drug dealers are killing each other and robbing each other, that is still X number of homicides and robberies during the year that the police investigates, while 0% of surveyed have been victims.

Why do you think that the survey wouldn't include drug dealers in the respondents?

Very few people are drug dealers, so unless you ask everybody, they will not be included.

Wouldn’t this be fixed by doing a proper, representative survey?

Insurance claims for stolen property and health stats around number of stabbings? Not perfect, but perhaps useful to validate UCR data.

How often do police departments have budget cuts? Union negotiated contracts usually do a good job of protecting at least the rank and file...

I would love to see any rigorously documented causation or even correlation between police budgets and crime rates.

> I would love to see any rigorously documented causation or even correlation between police budgets and crime rates.

Here's a bit of what we do have.

https://www.nytimes.com/2021/11/08/us/police-crime.html

https://abcnews.go.com/US/crimes-rise-battles-rage-police-fu...

A cursory watch of the "Hamsterdam" season of The Wire, or the "serial killer" one, will be enough to remove any faith in the accuracy of police stats. (Yes, it's fiction, but it's based on years of hardcore journalistic experience "on the crime beat" by show writers).

Reply on news.ycombinator.com