
Thursday, April 03, 2008 is new york safer? the other day at lunch, visitors from new york city proclaimed how safe their city is. new york city safe? apparently. and if we look at the statistics, new york city has a paltry per capita murder rate of 7.3. nyc doesn't even make it into the top 25, 30, or even 40. omaha and indanapolis rank far higher; even the "sunshine city" of st. petersburg florida comes in well above nyc. why does this ranking seem so wrong? perhaps because it is. as someone pointed out at lunch, this is the per capita rate. although your probability of getting shot in new york is much less according to the per capita rate, it would be much higher than sprawling indianapolis using a per mile rate. that is because new york is very small; all those murders are concentrated on a tiny island. i don't advocate always living life by statistics. however, if i am going to think about statistics, they'd better be the right ones. when i walked outside, i would ask myself: what is the probability of someone getting shot on this block right now? that is per mile rate, not the per capita rate (and neighborhood, time of day, etc. rates). in conclusion, nyc may be safer than it used to be. but you'll still have to carry your mace on your next visit. Labels: decision making permalink \| comments (10) \| 10 Comments: Umm. Were the NYC stats just for Manhattan? Manahattan is a small island (about 25 sq miles), but only a single borough. NYC has four other boroughs comprising 303.3 sq miles, which is only a smidgen smaller than Indianapolis at 372 sq miles. And your per mile rate seems rather dubious. How would you calculate the probability of the Bernoulli trial without having sample space defined on the occurrence/non-occurrence of the event? A rate of 20 per mile versus 200 per mile doesn't tell me very much about the probability of my being murdered. I'd be much more concerned by the former if the population is only 40 individuals within a mile, than the latter with a population of 400,000 individuals. By chris, at Fri Apr 04, 04:04:00 PM good point re: manhattan chris. i can't so far find the answer, and what, if 7.3 is not for just manhattan, that rate is. will report back. you're right i think about probabilities. though imo whatever that probability is, it doesn't matter to that one in x number of people who does get shot. By ns, at Fri Apr 04, 07:39:00 PM This post has been removed by the author. By WackedEcon, at Sat Apr 05, 02:02:00 PM At the close of 2007, Manhattan had 70 murders. Brooklyn, the most populous borough had 200, and so on. http://tinyurl.com/5vrz3c That's all out of 500 murders in entire city - not just Manhattan - for the year, the lowest since 1963. More importantly for the question of whether visitors to NYC need to worry: Just 100 of those 500 murders involved people who were not acquaintances with their assailant. That's the number that one would have to worry about if one was just visiting the city. The largest component of the other 400 were probably involved in the drug trade, in which case not getting involved in the NYC drug trade probably lowers your chance of getting shot quite a lot. As to your last line: Of course it doesn't matter to the person getting shot, but is that really the question? Or is the question, really, what is the likelihood of getting shot? By WackedEcon, at Sat Apr 05, 02:03:00 PM thanks for all of the useful info! i agree there are many more statistics of use than general murder rates, when considering how safe you are. but as to the question of "is the question, really, what is the likelihood of getting shot?" i am not sure, it is maybe a philosophical question. if i get shot, the probability of me getting shot was 100%, so the statistic that i was "safe" was of absolutely no use to me and had no predictive validity. i realize this is not a kosher view. By ns, at Tue Apr 08, 08:52:00 PM You having been shot doesn't change the underlying probability (from either a frequentist or Bayesian perspective); your having been shot is the realization of random variable. The probabilities we've been talking about would be conditional on not having been shot yet, in which case we ought to think of them as hazards (the instantaneous proby that x will happen right now \| on x not having happend prior to now). Of course all this would be conditional on other covariates (age, current location in NYC, involved in drug running, etc). Further (and albeit ironically, ns) once you've been shot, you will become a statistics. By chris, at Tue Apr 08, 11:41:00 PM yup. my inarticulate point was that in as much as an event is randomly-selected, the probabilities rather speak to a long-term accumulation of events than to one event itself. kind of. or something. By ns, at Sun Apr 13, 07:08:00 PM speaking of statistics: http://www.nytimes.com/2008/04/08/science/08tier.html?ex=1208318400&en=282eaa1e2cb5d56f&ei=5065&partner=MYWAY By ns, at Sun Apr 13, 07:10:00 PM The point of the post, as I gathered, was "is NYC safer" in terms of murder and you tried to argued that it was not, in fact, as safe as statistics suggest. However, it is, in fact, safer, in fact incredibly safe as compared to many other big cities. If the point of your post had been: "getting shot sucks", then I would've been happy to accept that as true. By WackedEcon, at Wed Apr 23, 02:26:00 PM whoa, easy now wacked econ. i try to maintain a friendly and cheerful environment here at FH, so please be nice. i have readily agreed about many points by commenters (commentors?) on this post, and in my last comment was simply continuing a conversation that arose on statistics in the comments section (that was the "point" was referring to. cheers, and be happy! n By ns, at Wed Apr 23, 02:51:00 PM Post a Comment the trackback URL for "is new york safer?" is: http://haloscan.com/tb/sullifred/6726474698222611775 trackbacks for this post temporarily listed here	recent posts a fruit by any other name... brain doping vs. steroids from the archives: age and decision making cleaning out my drafts box we like punishment health and discounting do alzheimer's patients remember social norms? why i haven't blogged in 400 years ants and economics o'rourke on wealth of nations categories well being social political personal neuroscience misc. science misc. psych links internet ideology economics drugs decision making search posts powered by free find blogrolling recommended books get posts via email enter your address: archives 03.2004 04.2004 06.2004 07.2004 08.2004 09.2004 10.2004 11.2004 12.2004 01.2005 02.2005 03.2005 04.2005 05.2005 06.2005 07.2005 08.2005 09.2005 10.2005 11.2005 12.2005 01.2006 02.2006 03.2006 04.2006 05.2006 06.2006 07.2006 08.2006 10.2006 11.2006 12.2006 01.2007 02.2007 07.2007 08.2007 01.2008 03.2008 04.2008 feeds & etc. feed options: main posts only current events only main posts and current events (also in rss) my posts at neuroeconomics.com creepy visitor map why no caps? the fox or the hedgehog? current events
design by me. all rights peacefully reserved, save where prohibited by law.

Thursday, April 03, 2008

is new york safer?

the other day at lunch, visitors from new york city proclaimed how safe their city is. new york city safe? apparently. and if we look at the statistics, new york city has a paltry per capita murder rate of 7.3. nyc doesn't even make it into the top 25, 30, or even 40. omaha and indanapolis rank far higher; even the "sunshine city" of st. petersburg florida comes in well above nyc.

why does this ranking seem so wrong? perhaps because it is. as someone pointed out at lunch, this is the per capita rate. although your probability of getting shot in new york is much less according to the per capita rate, it would be much higher than sprawling indianapolis using a per mile rate. that is because new york is very small; all those murders are concentrated on a tiny island.

i don't advocate always living life by statistics. however, if i am going to think about statistics, they'd better be the right ones. when i walked outside, i would ask myself: what is the probability of someone getting shot on this block right now? that is per mile rate, not the per capita rate (and neighborhood, time of day, etc. rates).

in conclusion, nyc may be safer than it used to be. but you'll still have to carry your mace on your next visit.

Labels: decision making

permalink | comments (10) |

10 Comments:

Umm. Were the NYC stats just for Manhattan? Manahattan is a small island (about 25 sq miles), but only a single borough. NYC has four other boroughs comprising 303.3 sq miles, which is only a smidgen smaller than Indianapolis at 372 sq miles. And your per mile rate seems rather dubious. How would you calculate the probability of the Bernoulli trial without having sample space defined on the occurrence/non-occurrence of the event? A rate of 20 per mile versus 200 per mile doesn't tell me very much about the probability of my being murdered. I'd be much more concerned by the former if the population is only 40 individuals within a mile, than the latter with a population of 400,000 individuals.

By chris, at Fri Apr 04, 04:04:00 PM

good point re: manhattan chris. i can't so far find the answer, and what, if 7.3 is not for just manhattan, that rate is. will report back.

you're right i think about probabilities. though imo whatever that probability is, it doesn't matter to that one in x number of people who does get shot.

By ns, at Fri Apr 04, 07:39:00 PM

This post has been removed by the author.

By WackedEcon, at Sat Apr 05, 02:02:00 PM

At the close of 2007, Manhattan had 70 murders. Brooklyn, the most populous borough had 200, and so on.

http://tinyurl.com/5vrz3c

That's all out of 500 murders in entire city - not just Manhattan - for the year, the lowest since 1963.

More importantly for the question of whether visitors to NYC need to worry: Just 100 of those 500 murders involved people who were not acquaintances with their assailant. That's the number that one would have to worry about if one was just visiting the city. The largest component of the other 400 were probably involved in the drug trade, in which case not getting involved in the NYC drug trade probably lowers your chance of getting shot quite a lot.

As to your last line: Of course it doesn't matter to the person getting shot, but is that really the question? Or is the question, really, what is the likelihood of getting shot?

By WackedEcon, at Sat Apr 05, 02:03:00 PM

thanks for all of the useful info! i agree there are many more statistics of use than general murder rates, when considering how safe you are. but as to the question of "is the question, really, what is the likelihood of getting shot?" i am not sure, it is maybe a philosophical question. if i get shot, the probability of me getting shot was 100%, so the statistic that i was "safe" was of absolutely no use to me and had no predictive validity. i realize this is not a kosher view.

By ns, at Tue Apr 08, 08:52:00 PM

You having been shot doesn't change the underlying probability (from either a frequentist or Bayesian perspective); your having been shot is the realization of random variable. The probabilities we've been talking about would be conditional on not having been shot yet, in which case we ought to think of them as hazards (the instantaneous proby that x will happen right now | on x not having happend prior to now). Of course all this would be conditional on other covariates (age, current location in NYC, involved in drug running, etc). Further (and albeit ironically, ns) once you've been shot, you will become a statistics.

By chris, at Tue Apr 08, 11:41:00 PM

yup. my inarticulate point was that in as much as an event is randomly-selected, the probabilities rather speak to a long-term accumulation of events than to one event itself. kind of. or something.

By ns, at Sun Apr 13, 07:08:00 PM

speaking of statistics:

http://www.nytimes.com/2008/04/08/science/08tier.html?ex=1208318400&en=282eaa1e2cb5d56f&ei=5065&partner=MYWAY

By ns, at Sun Apr 13, 07:10:00 PM

The point of the post, as I gathered, was "is NYC safer" in terms of murder and you tried to argued that it was not, in fact, as safe as statistics suggest.

However, it is, in fact, safer, in fact incredibly safe as compared to many other big cities.

If the point of your post had been: "getting shot sucks", then I would've been happy to accept that as true.

By WackedEcon, at Wed Apr 23, 02:26:00 PM

whoa, easy now wacked econ. i try to maintain a friendly and cheerful environment here at FH, so please be nice. i have readily agreed about many points by commenters (commentors?) on this post, and in my last comment was simply continuing a conversation that arose on statistics in the comments section (that was the "point" was referring to.

cheers, and be happy!
n

By ns, at Wed Apr 23, 02:51:00 PM

Post a Comment

the trackback URL for "is new york safer?" is: http://haloscan.com/tb/sullifred/6726474698222611775

trackbacks for this post temporarily listed here