Today’s Headlines

  • Cuomo and Bloomberg Make Federal Aid Push (Post, NewsAP)
  • Cuomo: $5B Would Get the MTA Back to Pre-Sandy Working Order (TransNat, SAS)
  • R Train Riders Packing 4 and 5 Trains (Post); South Ferry Won’t Reopen for Months (TransNat)
  • City to Provide Free Towing for Storm-Damaged Vehicles Left on Streets (Advance)
  • Gelinas: Mission Creep Debt Will Cripple the Port Authority for the Foreseeable Future (City Journal)
  • Ratner Withdraws Barclays Tax Suit (Bklyn Paper); AYR Puts Knicks-Nets in Perspective
  • Council to Regulate Pedicab Rates (Post); Vacca Wants Fare Decals Returned to Taxis (Post)
  • Limo Driver Charged With Manslaughter for Pinning, Dragging Passenger to Death (News 1, 2)
  • Manhattan Parking Garage Worker Crushed by Falling SUV (News)
  • Toddler Hospitalized After Being Thrown From Minivan in Woodhaven Collision (Post, DNA)
  • Kill Someone With Your Car, They Say, and NYPD Won’t Even Give You a Ticket (NY Mag)

More headlines at Streetsblog Capitol Hill

  • Larry Littlefield

    “Included in the report was the troubling news that in the previous year, from July 2011 to June 2012, traffic-related deaths had spiked upward by 23 percent, from 236 to 291.”

    Correlation is not causality, but what was the timing of the over-the-top media coverage claiming that NYC was engaged in a “war” against drivers, cheating them out of their ability to travel, in the wake of the Prospect Park West bike lane controversy?

    I believe that really got going in the winter of 2010 to 2011, just before the increase in fatalities.  And I did notice an increase in hostility while it was going on.  Drivers screaming at me because cyclists don’t stop at lights as I was waiting at a light, for example. 

    A little road rage and entitlement can go a long way toward explaining the increase.  Perhaps the added deaths can be counted as some as a success and not a failure.

  • jrab

    Figures for traffic deaths in New York City are sobering, but useless to consider in isolation. What is important is whether the city’s traffic-death rate is rising or falling RELATIVE to the national or state figures for traffic deaths.

    Unfortunately, the US Census has data through 2009 only. We will have to wait to see what more recent national data have to say.

  • Danny G

    @85211970d034887d032f8c319f70adbb:disqus National data is important, but NYC is in the major leagues. Statisticians should be looking at NYC relative to other international cities with populations of a few million or more.

  • Guest

    @85211970d034887d032f8c319f70adbb:disqus I’d argue that national data is mostly — but not completely — irrelevant without understanding the proximate causes of a rise or fall in fatality rates.  If traffic fatalities are going up nationwide and the rise can be definitively linked to a technological development such as smartphones or 3D movie glasses for drivers, then I’d pay attention to that info. (Indeed, one of JSK’s explanation for the jump here is smartphone use.)  Similarly, if rates are decreasing nationwide because of some new car design or safety standard, that’s worth noting.

    However, merely looking at national data and seeing if NYC fits the overall trend without this information contributes to the idea that traffic fatalities are as uncontrollable as the weather.  “Well, they’re rising everywhere, so I guess there’s not much NYC can do!”  As @36056f95783f8cfb512e9d49d4187ce6:disqus said, NYC is worlds apart from much of the rest of the country. Compare us to San Antonio and we look great. Compare us to Paris and we look terrible.

  • Anonymous

    I was expecting to find this NBC nightly news story cited somewhere on this site. 
    Hey, angry driver: Smile for the bike cam
    As commuting by bike becomes more popular, bikers are mounting small cameras on their bicycles to document what they say is aggressive driving. NBC’s Tom Costello reports.

  • jrab

    Danny G and Guest, I think you are both not fully grasping the concept of the chi-squared distribution and what it means. The null hypothesis (most simply, everything is the same) is that NYC is subject to the same trends of motor vehicle death as the US as a whole. Generally, as statisticians we choose tests to disprove the null hypothesis, which would mean that NYC has actually a different trend. (Obviously, New York City has a much lower vehicle death rate than the nation as a whole.) Once such a test has been run, we can investigate possible causes, which might have something to do with DOT street safety redesigns or public education campaigns, or with greater density in NYC, or with a different population in NYC, or with something altogether.
    If, however, a chi-squared distribution test does not disprove the null hypothesis, we cannot really assert that the DOT efforts have had any effect.

    To address Guest’s point, we cannot put the cart before the horse and try to “understand the proximate causes” without actually disproving the null hypothesis, that the year over year changes in New York’s traffic death rate are statistically significant in reference to the year-over-year changes in the national rate (not yet published for years more recent than 2009).
    To address Danny G’s point, the point of the chi-squared distribution is not to compare New York City to anywhere else. Rather it is to use a larger data set to see if the deviations we find in New York City data, specifically the year-over-year figures, are large enough to require an explanation other than chance

  • Joe R.

    @85211970d034887d032f8c319f70adbb:disqus I’m not even sure if looking at NYC data in isolation is all that useful. What we would really need to do is look not at the raw death/injury numbers, but at the numbers per mile traveled, or perhaps per trip. That gives us more useful information. If more pedestrians and cyclists are dying in the streets, it may simply be because more people are opting to walk or bike instead of drive or take public transit. The deaths per mile may actually be going down, which would affirm what has been done to make the streets safer. In many suburban and rural areas the number of pedestrian or cyclist deaths is very low or even zero. This isn’t because these places are so great for cyclists or pedestrians, but rather because nearly everyone drives.

    If I had to take a guess at future trends, I might say the number of deaths will continue to creep upwards for a while as yet more people opt to walk or bike. Once enough people are cycling or walking so as to take a substantial number of cars off the road, the numbers will trend downwards. In the meantime however, I’d bet good money the number of deaths per mile has been on a downward trend for the last 5 years, if not more. Sure, we need to do more to make the streets safer, but I think the per mile numbers would prove the value of what has already been done if they were available. In my opinion, it’s already far safer to walk or bike in NYC than at any time since maybe the 1940s or 1950s. Long term of course the goal should be zero deaths. With any luck we can get there within a decade.

  • Anonymous

    jrab, I’m not quite clear on your hypothesis.
    I think the basic question we want to answer is whether the number of observed traffic deaths in a given year represents a significant difference from the expected amount of traffic deaths, and if so, then we would want to know why.
    As such, we are more interested in the time series behavior of traffic deaths in NYC than we are with the cross sectional relationship of NYC traffic deaths to traffic deaths in other regions.

    Even if we accept the chi-sq distribution as a good model of the relationship between traffic deaths in different cities (which I’m not sure that it is), we still would not be able to use it to say anything about the trend of traffic deaths in NYC.
    I think Joe R is closer to the right track, where presumably we could construct a model to describe the time series of traffic deaths in various different regions, controlling for things like miles traveled, mode share, etc to normalize rates.  We still have the problem of trying to forecast a fairly rare event which probably has some weird tail-distribution properties.

    Anyway, 23% sounds like a big increase, but 55 deaths over 12 months is, unfortunately, likely to be statistical noise.

  • jrab

    J_12, excellent point. But looking closely at the “cross sectional relationship of NYC traffic deaths to traffic deaths in other regions” is a pretty basic way to evaluate the City DOT’s approach to traffic safety.