http://web-pblotter-v1-elb-1058384347.us-west-2.elb.amazonaws.com/SQLCrashData/
Keep in mind usage when you look at it. For instance, i'm willing to bet that, even though there are more crashes on Centre, that Baum is actually more dangerous, considering that there are probably way less people riding on it.
Also Negley - yeesh
kinda surprised at how few are on Smallman considering how popular of a route it is.
Keep in mind that this is only "reported crashes" ie ones that the police decided to write a police report over. That means that there pretty much had to be an injury for them to report it, as the thing that triggers a police report is an injury or a tow, and you don't tow a bicycle. Also, not every crash with an injury is even reported, even if the police show up. Technically any injury, even a bruise or scrape can be reported, but sometimes police will not write it up.
Anyway, pretty fascinating stuff and will hopefully help ya'll make informed route decisions.
erok
2016-10-06 16:30:20
Well my 2014 crash which included a police report is not listed. The fatal crash on Overbrook Blvd in Carrick is listed as injury unknown. And the fatal crash of the child down in Knoxville on Bausman Street (~2010?) isn't even listed.
Ya gotta love how accurate this police info is...
marko82
2016-10-06 16:52:44
True. I inquired about some of that in the past. it turns out that the child was not on a bike, but the media said that at first. not really sure how that happened, but penndot assured me it was not.
the overbrook one i'm not sure about. penndot claims that it's up to the local police to contact them and update the status. what can happen is that if a person leaves the seen alive, they put unknown on it, then if they die at the hospital, the police may not have followed up and wouldn't have updated it. it's messed up because you'd think there'd be a higher level of scrutiny and detail on this stuff when there is a fatality or serious injury.
When i first started digging into this, they had james price listed as a "pedestrian on skates" i informed the police dept about it, so i'm glad that it got updated at least. but penndot claims that it's on the police to update
erok
2016-10-06 16:59:56
Re fatality on Bausman: They assured me that he was just crossing the street, and were not sure where the bike story came from. this video kind of confirms it a bit
https://www.youtube.com/watch?v=2GJ6xgs-iGM
erok
2016-10-06 17:05:46
as for why yours didn't show up...no clue
erok
2016-10-06 17:08:44
one thing that i've noticed is that coding is not consistent, like the james price issue. My guess is that they probably coded yours as a pedestrian
erok
2016-10-06 17:11:12
I don't see either of my crashes recorded.
This is a bit worrisome. There are issues of incorrect data which will happen (but you'd want the issuer to provide an estimate). And there's clearly issues of omission which I think is more serious. These data end up as input into policy decisions: the current data will suggest that bikes crashes are less of a problem than they actually are.
This systematic bias puts cyclists at a serious disadvantage: threats will not be taken seriously ("sorry, there's not enough data to support the need for x"). This is not at all good.
Can BikePgh bring this up with the proper authorities?
ahlir
2016-10-06 18:44:39
Agreed. We have brought this up. it's a serious issue, and we have been in talks with the PPD to take it more seriously and get better at it. with that said, the city has only recently been taking data very seriously as well, so there is more than just us that are pushing for better crash reporting.
erok
2016-10-06 19:19:20
I am sure that Erok's information is very solid. It's easy to miss an accident or two.
If interested, Erok, I have a hard copy of all the self-reported accident "stories" from the old BP Bike Crash map.
I also have the data from the PG article from a few years back.
I assume that this information is either a) already available to you, or b) of lesser interest than the PennDOT data that you obtained, and you do not want to taint the "scientific" data.
However, if you are interested in having either, or both, as a supplemental data record, just let me know, and I'll shoot it over to you.
swalfoort
2016-10-07 08:54:49
A technical problem with that map: the icon locations drift as you zoom out. The dots appear to have their origin at the bottom while the stars have their origin at the upper left. If you zoom out a lot you get an inaccurate picture.
If these data were integrated with Strava's statistics on bicycle traffic, you could get a better estimate of the risk of each street. ... And if you had Google's statistics on cell phone locations, you could look at correlation between cyclist crash risk and car speed...
paulheckbert
2016-10-07 09:22:15
that original link no longer works use this one.
http://codeforpittsburgh.github.io/BikeCrashMap/
erok
2016-11-08 15:10:56
