True, CompStat is not at all "widely credited" with contributing to the nationwide drop in crime. Even its contribution to NYC's drop is debatable: https://en.wikipedia.org/wiki/CompStat#Critique
"Since Compstat was introduced, crime rates in New York City have dropped dramatically. From 1993 to 1995, the total crime rate declined 27.44 percent across the city."
National crime rates dropped as well in that period but nowhere near 27%.
> how evolutions in data-led policing strategies helped New York City reduce annual murder numbers from 2,245 in 1990 to just 292 in 2017
This is a highly contested claim, because this happened everywhere in the US across the same time period, whether or not they were using CompStat or CompStat inspired strategies.
Oops, didn’t notice the first link was for New York State. The CompStat link includes longer-term trends, though, which is what I was referring to.
Crime is down overall, and I suspect that if you controlled for the murder rate increase in all US cities over the past two years it might be flat or down in NYC as well.
Crime rates nationally dropped also during the same time - and not all areas had the same analytics. It's really not clear if the reductions were from the NYC stats or in particular "agressive" law enforcement.
Maybe. Many pundits have been talking about NYPD's use of COMSTAT, stop-and-frisk and fighting the "broken windows" syndrome as being leading causes of the drop in crime in NYC. On closer inspection the crime appears to have been falling anyway for a number of possible reasons, some as simple as a change in demographics.
Now we have perhaps something of a controlled experiment going on regarding the alleged root cause of the drop in crime in NYC (and perhaps nationwide).
When averaging results over a whole city, you've eliminated most sources of noise. But if you average neighborhood by neighborhood, then you have 20x smaller samples, and on top of that, 20x more chances to p-hack. So it's unsurprising to me that there exists some neighborhood where crime rates dropped significantly in one year.
I think GP's comment would be stronger if they had given absolute numbers instead of "67% decrease".
No you can not just look at murders and declare that over all crime is down because murder is down.
Looking at cities that do not have the same reporting problem that these large cities with a certain political bent you can see that murder does not track with non-violent crime aka property crime and only has a loose correlation to other violent crimes.
it is entirely possible for Theft, robbery, Rape, etc to be increase while murder is decreasing.
Further according to U.S. Department of Justice’s Bureau of Justice Statistics issued in March 2023, between 34% and 58% of violent crimes were reported to police. For property crimes, the percentage reported to police was between 28% and 44%.[1] It may seem add on only 34% of violent crime would be reported, but we have seen many reports in recent years of police treating victims like criminals in the first place so many people will just move on and never involve the police even with violent crime.
Then there is the reported 90,000+ package thefts [2] that occur that NYPD does not put on their comstat figures they release as these are considered minior offenses, but have a real impact on daily life (aka broken windows)
I’m not trying to be confrontational, but I do need to say something on this comment and hopes for some good discussion.
How much of that is due to actual drops in crime levels and how much of it is due to decriminalization of offenses and a the resultant drop in crimes reported?
Both the Uniform Crime Reporting Program from the FBI and the Bureau of Justice Statistics’ National Crime Victimization Survey have methodological have problems. There isn’t even a consensus between when about what constitutes criminal activity. And in some jurisdictions, only incarceration worthy crimes are considered ‘crime,’ where as in others fined infractions like speeding in a vehicle count. To say nothing of the tendency of crime data to suddenly regress towards the mean, which has been a problem for researchers.
I am wary to accept short term crime statistics, because it’s so politicized. There’s a great discussion here, though it’s a tad dated:
Sure, the statistics are not very good, but we don't have anything better. We have some evidence that property crime has gone down relative to other parts of the country (and much better evidence for homicide, which doesn't have the same under-reporting risk), and no evidence (as far as I know) for the opposite.
I listened to that a few years ago. I don't believe it makes exactly the claim as your summary.
In discussions about New York there is a popular myth that policing caused the crime rate drop, and I think that is what the podcast was challenging. Not necessarily that there wasn't a drop at all. There was across the entire country. Not for the reasons some police departments would like to believe.
cannot agree more. in my opinion, regardless of the statistics drop in crime is primarily attributed to exuberant increase of the cost of living especially in the historically impoverished neighborhoods from the 1980s and on. (i.e. Bushwick, Williamsburg, Bedford-Stuyvesant etc...)
> The contribution of such deterrence measures (the "stick") offers more explanation for the decline in New York City crime than the improvement in the economy, the authors conclude. Between 1990 and 1999, homicide dropped 73 percent, burglary 66 percent, assault 40 percent, robbery 67 percent, and vehicle hoists 73 percent. The authors' model manages to explain between 33 and 86 percent of those declines
I'm not saying crime is going up. I'm saying the stats are useless. It could be going down, and they're still useless.
Yes, I am talking about NYC. I'm sure every other city is a relative statistical utopia. But I have no information on those cities, given I don't live there, except for the official stats.
They could be shaving off exactly 10% every year, and the trend lines are fine. Or they could shave off enough to make crime go down slightly every year, which is what I'd do if I were a chief who had to answer to a mayor, and didn't really care about statistical fidelity.
Just because you're looking at a chart, doesn't mean the chart reflects reality. And it doesn't matter if you don't have a nice other chart to look at. Sometimes you can't answer questions accurately from your chair 500 miles away.
It's slightly more than just interesting. It's been a fairly long-held belief that, at least in New York City, their continued and constant drop in crime was largely attributable to the fall in popularity of crack and cocaine usage.
I'm not sure why he's been downvoted to gray. I don't know that his ideas extend well to areas outside of NYC, but there's at least one (fairly large) data point lending credence to his notions.
Extrapolating from a single neighbourhood to an entire city is not valid reasoning.
Perhaps the neighbourhood was especially amenable to the program, perhaps the crime rate failed to stay low, perhaps some extra-judicial phenomena struck that particular area (say a new factory opening up that boosted the local economy significantly). Or coming at it from the other angle, who's to say New York hasn't had one or two of its neighbourhoods experience extreme marked drops in crime year-over-year too? The 75% over a decade figure is an aggregate for the entire city, and surely it's going to have happened faster in some areas and slower in others.
Only by comparing apples to apples, or in this case cities to cities, and not just for a single year but over many years can we reach valid conclusions. And in that case we find the VRU approach in Glasgow failed to out perform the null hypothesis.
I grew up in NYC during the high-crime Lindsay/Koch years. I would apply some skepticism to crime statistics derived from police reports. While I don't doubt that violent crime is way down, it's also true that people tend not to bother to report many crimes, because the police don't seem to be interested. While I'm sure that most murders would end up in the official statistics, I know that plenty of assaults, etc. are just never reported.
Are you claiming that New Yorkers report crimes less frequently than others? It'd be great to have a source for that; absent that, there's no reason to believe that any sample error in the NYC crime data isn't similarly reflected in all crime data.
It also doesn't seem to lower the crime rate. [...] There were more social programs and money pumped into that city than any other, yet the murder rate is #2 in the US.
This should be pretty obvious to most HN readers, but I'm reminding people anyway. "Correlation does not imply causation".
reply