Two more “missing” hit-and-run fatalities

First of all: WAY TOO MANY HIT AND RUNS. Fully ONE THIRD of all fatal Bike-MV crashes in 2016 had the driver fleeing the scene, a felony.

 

“Missing” refers to the database has these listed as not a hit-and-run, yet they are missing virtually all identifying data, except the victim’s demographics. The only thing I can imagine is they are hit and run, and are mis-coded.

In reconciling  asdm data for 2016 these two fatalities appeared. Given the paucity of data and the profusion of UNKNOWNs these two appear to simply have the hit-and-run flag mis-coded?? One has absolutely no location information, i.e. no streets, and no Lat/long, the other lists a location.

It’s hard to understand how the data can be so shabby. I don’t understand how this can happen. ADOT should reject such reports. Crappy data.

Anyway, that would mean that “official” counts of hit and run fatalities may be undercounted because of poor reporting practices. There were also two similar to this last year (2015), for other suspicious incidents, search on the tag missing-hit-and-run); in other words, this isn’t just a one-off thing.

In 2016, according to the official ASDM statistics,  8 (of 30  = 27%) fatal bicyclist incidents were hit-and-run; if two are in fact missing the actual percentage would be 33%.

Goodyear 4/1/2016 12:30pm Incident

  • Location: No Data
  • Officer / Extended NCIC: Goodyear PD
  • Victim: 65/M
  • Driver: 115/Unspecified gender
  • Vehicle: Unknown Make and bodystyle
  • Hit and Run : No
  • Unit Action / Direction / etc (both units): UNKNOWN

Maricopa Co. 11/13/2016 00:00 Incident

  • Location: Ellsworth Rd & Pueblo Ave
  • Officer / Extended NCIC: Maricopa Co Sheriff (MSCO)
  • Victim: 57/F
  • Driver: 115/Unspecified gender
  • Vehicle: Unknown Make and bodystyle
  • Hit and Run : No
  • Unit Action / Direction / etc (both units): UNKNOWN

Reverse googling the 2nd incident reveals this message from the MSCO public info officer’s twitter, on 11/13/2016 10:06pm; with no further information.

How many other “missing” Hit-and-Runs are there?

I don’t know but here’s some guesses… Any incident where a Driver’s age is 115 (or 255 or 254? p.s. the 255 could involve “DRIVERLESS”, i.e. parked, cars?) AND the H/R flag is not set is highly suspicious.  Below are counts just for Bike-MV and Ped-MV:

H/R Bike c_Age count(*)
0 115 15
0 255 6
1 115 37
1 255 139
H/R Ped c_Age count(*)
0 115 20
0 254 1
0 255 60
1 115 66
1 255 240

.

Historical curiosity; the age is 115 is for the year 2016, it’s gone up by a year every year I’ve looked at, e.g. in 2015 the magic age was 114, in 2014 it was 113, and so on.

So out of ~ 3,500 crashes (ped + bike), there are around 100 that have some funky data. Is that just random mistakes?? seems high. I have yet to work on a query that looks at vehicle-vehicle crashes, here the ped/bike one:

SELECT
i.HitAndRunFlag "H/R Bike", p_car.Age c_Age, count(*)
FROM ((((
2016_incident AS i JOIN 2016_person AS p_bike ON i.IncidentID = p_bike.IncidentID)
 JOIN 2016_unit AS u_bike ON p_bike.UnitID = u_bike.UnitID) 
 JOIN 2016_person AS p_car ON i.IncidentID = p_car.IncidentID)
 JOIN 2016_unit AS u_car ON p_car.UnitID = u_car.UnitID )
WHERE p_bike.ePersonType = 'PEDALCYCLIST' AND p_car.ePersonType='DRIVER' AND p_car.Age > 100 
GROUP BY 1,2;

.

Hit and Run Flag for All Incidents

Here’s what i put together to try and help make sense, if possible, from all the data, not just ped or bike…

In 2016,  about 12% ( 14,912 of 126,845) of all reported crashes were hit and run according to the Incident table. There were a total of around 250,000 “units” involved in crashes (i.e. most crashes involve 2 units; some just one, and some more than 2). The last query breaks down the Ages of the driver, and whether or not the incident was a hit and run in the incident table, and the status of the hit and run flag in the unit table. Odd driver ages (either 115 or 254, 255) are suspicious if the unit doesn’t have hit and run.

Here are the suspicious incidents:

H/R incident H/R unit unit type Age COUNT
1 0 DRIVER 115 120
1 0 DRIVER 255 257

.

SELECT count(*) FROM 2016_incident WHERE HitAndRunFlag=1;
SELECT count(*) FROM 2016_incident;
SELECT eUnitType, count(*) FROM 2016_unit GROUP BY 1;
SELECT
i.HitAndRunFlag "H/R incident",u.HitAndRunFlag "H/R unit", u.eUnitType, p.Age, count(*)
FROM ((
2016_incident AS i JOIN 2016_unit AS u ON i.IncidentID = u.IncidentID)
 JOIN 2016_person AS p ON (u.UnitID = p.UnitID) AND (p.ePersonType='DRIVER') )
GROUP BY 1,2,3,4;

.

2 thoughts on “Two more “missing” hit-and-run fatalities”

  1. Early in my career, I worked with this kind of “accident” (collision) data for years. I’ve examined the details of thousands of collision reports. In those days, a paper report could be pulled to compare with data entered. Even the original paper reports would contain errors. I’m not in the least surprised the data, in this case, is inconsistent or miscoded. Anyone relying on the data, should understand the potential for errors.

    Most likely, there are few if any checks or auditing practices occurring in the agencies when reports are “filed” or input into the data system. Anyone that has worked much with data knows you have to employ auditing and checking protocols – as errors and omissions in the data will always occur. Auditing and checking must be utilized to catch the inevitable errors.

  2. To validate/verify that the incident’s HitAndRunFlag has at least one unit’s HitRunFlag set do the following query; “Having” is like WHERE but runs after “GROUP BY”/aggregation (max/sum/etc):

    SELECT IncidentID, i.HitAndRunFlag,  MAX(u.HitAndRunFlag) FROM 2016_incident i JOIN 2016_unit u USING (IncidentID) GROUP BY 1,2 HAVING i.HitAndRunFlag!=MAX(u.HitAndRunFlag);

Leave a Reply

Your email address will not be published. Required fields are marked *