It's a question of framing and being able to put a label on the report. Since no label exists for the (chinatown) area, the report states that Broadway and Third avenue are the most dangerous roads, when in fact other areas have much higher density of fatalities.

http://www.tstc.org/reports/danger08/manhattan.pdf

The report is vehicle centric in that it skews focus from most dangerous areas (for pedestrians) to favor the longest continuous roadways from a drivers point of view.

2.0's out of beta now -- the performance has been improved, along with some other usability improvements and bug fixes.

The feature you want is coming in version 2.1, to be released later this year.

The same sort of issues occur with interpreting the data on CrashStat.org (we just released a new version yesterday, by the way). Although certain intersections may seem more dangerous than others in terms of raw numbers of fatalities and severe injuries, what one really needs is a denominator -- pedestrian and motor vehicle volume at each intersection. Unfortunately such information is not available.

Where's the new version? It still says "Crashstat 2.0." Zooming in on an intersection or zip code using the Google Maps search functionality would definitely be helpful.

It seems like most of the vehicle and pedestrian volume data should probably be computerized now. What's preventing the DOT from simply releasing the raw data on the web? Is there any justification for it other than bureaucratic secrecy?

On the other hand, Michelle and I figured it would distort the analysis in the opposite direction, making short roads appear more dangerous than they are.

I think that would only happen if you compare averages for the streets. If you talk about specific miles instead, such as "the easternmost mile of 125th Street," then it doesn't say anything about the rest of the road.

For what it's worth, in suburban Long Island and New Jersey, fatalities do appear to be clustered along identifiable roads. Yes they're in areas with lots of pedestrians, but the rest of the story is the speed that cars pick up between lights and, in many cases, inadequate pedestrian crossings and sidewalks.

That definitely makes a lot of sense. Essentially, it's the State DOT and local governments wanting to push cars through at any cost, even in places that attract a lot of pedestrians.

You're right, fatalities by route mile would be good. On the other hand, Michelle and I figured it would distort the analysis in the opposite direction, making short roads appear more dangerous than they are.

In other words, a three-block long street with one or two fatalities would probably top the list.

There's a compromise here somewhere but it's likely to be a bit more complicated (for example, breaking it down by classes of road or excluding roads shorter than a certain distance).

For what it's worth, in suburban Long Island and New Jersey, fatalities do appear to be clustered along identifiable roads. Yes they're in areas with lots of pedestrians, but the rest of the story is the speed that cars pick up between lights and, in many cases, inadequate pedestrian crossings and sidewalks.

So, I guess this is pretty obvious, but it seems like the combination of pedestrians, fast car traffic and inadequate pedestrian infrastructure produce suburban danger spots.

Definitely good points though.

Westchester suffered a high number of pedestrian fatalities - 23, but the deaths were distributed throughout the county so that none of Westchester's roads ranked as "most dangerous."

Good comment about running the analysis by route-mile...we've been playing around with the data to try to do just that.

I can understand why they chose to focus on roads, but it does complicate the analysis, because all other things being equal, a long road will have more deaths. I would be interested in seeing the data analyzed from a route-mile perspective, the "ten deadliest miles of road." For example, it looks like five of the deaths on Woodhaven/Cross Bay Boulevard were concentrated in Ozone Park, where there are more pedestrians. Similarly, there are four deaths on 125th Street between Park Avenue and the Triboro Bridge. On Hempstead Turnpike/Fulton Avenue, four deaths occurred in Hempstead, two in West Hempstead and two in East Meadow.

Given all the carnage in Manhattan and Queens, I'm surprised that Hudson County seem so safe.

Also, where's Westchester?