Collision Manner

[Warning/correction not yet made: in the table below where it says MV-only, that’s not quite correct, it’s really incidents where no pedalscylists are involved. The MV-only calculations should also exclude pedestrians; this can be accomplished by saying u.eUnitType LIKE ‘PED%’  In the query below i corrected it but didn’t correct the table; the percentages don’t really change since there are relatively few ped crashes ]

Here is a breakdown of Collision Manner, and rates, for MV collisions (i.e. one or more MV, and not involving a ped or bicyclists) compared to bike-MV collisions.

The megatrends are that rear end collisions are, by far, the predominant manner of collision for MVonly crashes; wheres for bike-MV crashes this manner is quite infrequent — almost twelve times more frequent. For bike-MV crashes, the predominant manner is angle, i.e. so called “turning and crossing” movements (although left turn is broken out as a separate manner).

The 38% rate of REAR END crashes for MVs actually understates the rate — if you back out the number of SINGLE VEHICLE crashes; you see that nearly half of all multi-car collisions are REAR END(!). 38,499/(101,055 – 18,647) = 47%. Inattention? Does this mean that motorists actually are more attentive to same-direction traffic when it’s a bicyclist, compared to another motorist? hmmm.

+------------------------------+-------+--------+------+--------+-------------+
| eCollisionManner             | MVonly| MVrate |bikeMV|bikerate|MV:bike ratio|
+------------------------------+-------+--------+------+--------+-------------+
| REAR_TO_REAR                 |   287 | 0.0028 | NULL |   NULL |        NULL |
| UNKNOWN_99                   |   859 | 0.0085 |   40 | 0.0189 |  0.44973545 |
| REAR_TO_SIDE                 |   895 | 0.0089 |   11 | 0.0052 |  1.71153846 |
| SIDESWIPE_OPPOSITE_DIRECTION |  1244 | 0.0123 |   46 | 0.0217 |  0.56682028 |
| HEAD_ON                      |  1438 | 0.0142 |   45 | 0.0212 |  0.66981132 |
| OTHER_97                     |  3160 | 0.0313 |  404 | 0.1905 |  0.16430446 |
| SIDESWIPE_SAME_DIRECTION     | 10727 | 0.1062 |  124 | 0.0585 |  1.81538462 |
| LEFT_TURN                    | 11888 | 0.1176 |  189 | 0.0891 |  1.31986532 |
| ANGLE_FRONT_TO_SIDE          | 13411 | 0.1327 | 1194 | 0.5629 |  0.23574347 |
| SINGLE_VEHICLE               | 18647 | 0.1845 | NULL |   NULL |        NULL |
| REAR_END                     | 38499 | 0.3810 |   68 | 0.0321 | 11.86915888 |
+------------------------------+-------+--------+------+--------+-------------+
total num of MVonly crashes = 101,055.  total num of bike-MV crashes = 2,121
source: 2012 ASDM

It seems like it would be useful to plug in injury severity; and check for consistency with previous data, e.g. Cross and Fisher; to see relative injury rates — e.g. we might expect to see rear end for bicyclists (which are low frequency) result in high relative rates of injury.

ALL 18 fatalities in 2012 were OTHER (as were all 23 2011 fatalities); whereas in 2009, and 2010 there was what you might expect a variety of manners. It seems to suggest strongly that someone somewhere is making a concerted effort to mask this data.(?? why?). A spot-check of 3 fatal ACRs I have for 2011 and 2012 show they all have ANGLE. Somebody is diddling with the data. In any event; this field has no useful information in it anymore with respect to fatalities.

[sidenote on PEDESTRIAN. The same queries can be run with PEDESTRIAN in place of PEDALCYCLIST. There is some sort of scrubbing going on there as well. In 2012 all 130 ped fatals were OTHER except for one HEAD_ON. In 2011 all 146 were OTHER. Older data is more mixed, but curiously 2010 is more mixed than 2009 was. Also if you look at fatal incidents, without regard to whether it was MVonly, MV-ped, or MV-bike, you see a broad mix of CollisionManner; and that the OTHER is comprised mostly of fatal peds + bicyclists. Incidentally, the SINGLE_VEHICLE category is by far the largest when looking at all fatal incidents, at 316 is like 1/3 of the total]

Some clouds in the data

[note: data fiddling / fudgery on manner of collision lands here at #clouds. That refers to the habit of crash reports to get somehow and inexplicably (by whom? by ADOT? no one will say) to summarily override the manner of collision from, say, a rear-end to OTHER. This particular field, nearly without exception, gets overridden for bicyclist and ped fatalities.  MESSING AROUND WITH THE DATA MAKES IT MORE DIFFICULT TO ANALYZE]

The crash manual (see adot-traffic-collision-database “reference document” p.37) has some confusing/contradictory verbiage associated with this. It is Block 17 on the ACR form; the “MANNER OF CRASH IMPACT- Identifies the manner in which two motor vehicles in transport initially came together”.  Strictly speaking, there is no possible correct answer as to any bike-MV (or ped-MV) crash; since those crashes are not two motor vehicle. It appears that at least most crash reports treat a bike, for the purposes of this block anyway, as one motor vehicle in transport.

I had a brief email exchange with Tom Bragan @dot.gov over the equivalent field in the mmucc (actually now C9 in the 4th ed) , where pretty much the same confusion/contradiction exists. He said he would take that under consideration for the next rev of their documentation (the 5th edition). By the way, the mmucc definitions, compared to AZ, lack a SINGLE VEHICLE, and LEFT TURN (which is simply a subset of ANGLE).

It seems like the best thing to do to clear up the discrepancy is to simply expand the definition to “two motor vehicles in transport, or at least one motor vehicle in transport and one other traffic unit in transport”.  And adjust each manner’s definition accordingly.

Queries

Make the big table to compare MV(only) manners and rates to MV-bike manners and rates:

SELECT 0 INTO @b; SELECT 0 INTO @a;
SELECT a.eCollisionManner, @a, a.c, a.r, @b, b.c, b.r, a.r/b.r, b.r/a.r FROM
 (SELECT eCollisionManner, count(1) c, @a, count(1)/@a r FROM
  (SELECT *, @a:=1+@a FROM 2012_incident i WHERE NOT EXISTS
    (SELECT 1 FROM 2012_unit u WHERE i.IncidentID=u.IncidentID AND u.eUnitType LIKE 'PED%')) x GROUP BY 1) a
LEFT OUTER JOIN (SELECT eCollisionManner, count(1) c, @b, count(1)/@b r FROM
 (SELECT *, @b:=1+@b FROM 2012_incident i WHERE EXISTS
   (SELECT 1 FROM 2012_unit u WHERE i.IncidentID=u.IncidentID AND u.eUnitType='PEDALCYCLIST')) y GROUP BY 1) b
ON a.eCollisionManner=b.eCollisionManner GROUP BY a.eCollisionManner ORDER BY 3;
### Note: the clause LIKE 'PED%' will exclude both pedalcyclists and pedestrians


Select by manner; differentiate by UnitNumber and InjurySeverity:

SELECT eCollisionManner,count(1) FROM 2012_incident i WHERE EXISTS 
 (SELECT 1 FROM 2012_unit u WHERE u.IncidentID=i.IncidentID AND 
   u.eUnitType IN ('PEDALCYCLIST') AND u.UnitNumber IN (1,2) ) 
  AND InjurySeverity IN (1,2,3,4,5)  GROUP BY 1 ORDER BY 1 ASC;

 

5 thoughts on “Collision Manner”

  1. Odd crashes in the 2012 ASDM dataset

    The vast majority of bicyclist crashes involve exactly one bicyclist and one motor vehicle (remember, crashes involving no MV are excluded from the dataset, by definition)

    IncidentId 2596875
    NINE units (8 cyclists and 1 driver); REAR END
    unit 1 was a cyclist; all the cyclists got a SPEED TOO FAST violation. The driver got NO IMPROPER… i.e. 8 cyclists ran into the back end of the car. The car’s UnitAction: STOPPED_IN_TRAFFICWAY
    location SR 187
    officerNcic / extendedNcic : 799 / 1100; which is DPS/Pinal Co. Sheriff’s Office

    Other large number of traffic unit crashes:

    These had 4 units: 2604722 | OTHER_97, 2629484 | OTHER_97, 2631136 | REAR_END (this one was the infamous GPS lady that ran into 3 cyclists from behind)

    There were another 50 or so that had 3 units; about half had 2 MVs and one bicyclist, and the other half was 2 bicyclists and one MV. See complete number of units breakdown at http://azbikelaw.org/contrib/asdm/bsap-data.txt [15]

    HISTORICAL: the largest number of bicyclist units involved in a single incident in prior years was:
    2009, 7 (IncidentId 2346120; SIDESWIPE_SAME_DIRECTION, way outside Tucson, motorist at fault; worst injury is POSSIBLE)
    2010, 3 (three instances);
    2011, 4 (IncidentId 2533594; OTHER, far north AZ, rt 89 driver was overtaking, driver faulted; one INCAP injury)

  2. —– Forwarded Message —–
    From: Ed Beighe
    To: “Tom.Bragan@dot.gov”
    Sent: Monday, July 29, 2013 12:10 PM
    Subject: Re: [Message] bikes, peds and C8. Manner of Crash / Collision Impact

    Hey tom,
    thanks.

    Nothing to do with the question about bikes/peds: I misspoke earlier when I said AZ matches what you have in mmucc for Manner of Collision:
    AZ’s ANGLE excludes left turn, and adds a LEFT TURN choice. We also have and a SINGLE VEHICLE, which is inconsistent with the definition of the field (“the manner in which two motor vehicles…”)

    From: “Tom.Bragan@dot.gov”
    To: Ed Beighe
    Sent: Monday, July 29, 2013 5:34 AM
    Subject: RE: [Message] bikes, peds and C8. Manner of Crash / Collision Impact

    Ed – I agree with you and will include your suggestion in my 5th Edition possible changes.
    Thank you,
    Tom

    From: Ed Beighe
    Sent: Friday, July 26, 2013 3:57 PM
    To: Bragan, Tom (NHTSA)
    Cc: Justin Pryzby
    Subject: Re: [Message] bikes, peds and C8. Manner of Crash / Collision Impact

    Hey Tom,
    thanks for the reply. Your answer, though definitive, seems to me to needlessly get rid of important/useful data, especially in the case of bike-MV.
    Can you tell me where /how such decisions get made?

    It seems like a better way to handle it is to clarify or define a bike to be allowed to “count” as one motor vehicle in transport, at least for the purposes of this data field. This seems to be the way, de facto, it is most often coded.

    Below(in case you are interested) is the breakdown of bike-MV collisions in Arizona last year… AZ uses the same definitions as the mmucc
    Thanks!
    +—————–+——————————+———-+
    | CollisionManner | eCollisionManner | count(1) |
    +—————–+——————————+———-+
    | 2 | ANGLE_FRONT_TO_SIDE | 1194 |
    | 3 | LEFT_TURN | 189 |
    | 4 | REAR_END | 68 |
    | 5 | HEAD_ON | 45 |
    | 6 | SIDESWIPE_SAME_DIRECTION | 124 |
    | 7 | SIDESWIPE_OPPOSITE_DIRECTION | 46 |
    | 8 | REAR_TO_SIDE | 11 |
    | 97 | OTHER_97 | 404 |
    | 99 | UNKNOWN_99 | 40 |
    +—————–+——————————+———-+

    mysql> SELECT CollisionManner,eCollisionManner, count(1) FROM 2012_incident i WHERE EXISTS (SELECT 1 FROM 2012_unit u WHERE i.IncidentID=u.IncidentID AND eUnitType=’PEDALCYCLIST’) GROUP BY 1 ORDER BY 1 ;

    From: “Tom.Bragan@dot.gov”
    To: Ed Beighe
    Sent: Friday, July 26, 2013 12:37 PM
    Subject: RE: [Message] bikes, peds and C8. Manner of Crash / Collision Impact

    The correct answer for the two scenarios you mention, which is not listed in the Guideline, is “Not Applicable” — that’s a correction that needs to be made.

    Thanks,
    Tom Bragan

    —–Original Message—–
    From: registration@mmucc.us [mailto:registration@mmucc.us]
    Sent: Friday, July 26, 2013 3:01 PM
    To: Bondy, Nancy (NHTSA); Bragan, Tom (NHTSA); O’Neil, Morrie CTR (NHTSA); joncjordan@gmail.com
    Subject: [Message] bikes, peds and C8. Manner of Crash / Collision Impact

    Ed Beighe sent a message using the contact form at http://www.mmucc.us/contact.
    http://www.mmucc.us/data-elements/crash-data-elements/c8-manner-crash-collision-impact

    What is the correct answer for this data field when the crash is a bike-MV or ped-MV crash?
    (that is to say: the crash is not between “two motor vehicles in transport”)

  3. data fiddling and collisionManner. Below might be a little hard to read; but it’s 4 years (2009-2012) of bicyclists fatals broken down by CollisionManner. I seriously doubt investigators all across arizona suddenly (commencing with 2011) started coding 100% of bicyclist fatals as “OTHER” all by themselves.
    In any event, i think this confusion can and should be cleared up simply by tweaking the instructions to treat a bicycle as a MV at least for the purposes of this data field. I wonder who authored that comment in the CoP bike plan? I feels like they possibly don’t understand what’s going on(?) and it feels like somebody from adot isn’t telling the whole story.

    There’s not a clear pattern if i look at other types of injuries, e.g. serious injuries (an injury severity of 4 on a 1 to 5 scale).

    eCollisionManner 2009:
    ANGLE_FRONT_TO_SIDE 6
    LEFT_TURN 1
    REAR_END 9
    HEAD_ON 1
    SIDESWIPE_SAME_DIRECTION 3
    OTHER 5

    eCollisionManner 2010:
    ANGLE_FRONT_TO_SIDE 6
    LEFT_TURN 2
    REAR_END 2
    HEAD_ON 2
    OTHER 7

    eCollisionManner 2011:
    OTHER 23

    eCollisionManner 2012:
    OTHER 18

    SELECT eCollisionManner,count(1) FROM 2012_incident i WHERE EXISTS (SELECT 1 FROM 2012_unit u WHERE u.IncidentID=i.IncidentID AND u.eUnitType IN ('PEDALCYCLIST') ) AND InjurySeverity IN (5) GROUP BY 1 ORDER BY 1 ASC;
    Oh, and by the way, the pattern for PEDS is that almost all (say all but 5 or 6 out of 120/yr), for all 4 years have been OTHER. So the pattern is different for PEDS than for bicyclists.

    This appeared in the Phoenix Bicycle Master Plan (draft March 2014)… I’m not sure i agree with the bit about “change in procedures” but in any event, somebody else has noticed the problem:
    “The current crash report form and police officer reporting may not provide enough detail to accurately describe bicyclist crashes. Upon review of fatal and serious injury (K and A) crashes involving bicyclists by manner of collision, approximately 50% are coded under ‘other’ or ‘single vehicle.’ This may indicate a greater need for police officer training on how to accurately fill out crash reports that involve bicyclists. It would also be desirable to include more types of bicycle-involved crashes in the ALISS crash database to better monitor the safety of the road network for bicyclists. This would require a change in the ADOT reporting procedures”

  4. Someone wondered about motorcycle collisions; rather than bicyclist REAR-END collision.
    The queries are below, there were a total of 3167 MC collisions (one or more MCs involved).
    647 (or 20% of the total of 3167 incidents) of these were REAREND, so that’s somewhat lower than the 37% overall for all motor vehicles. But that’s in total; In a REAREND, fault is pretty clear — if you filter on MC not at fault (i.e. when a MC is unitNumber 1) you see 302 or 1203 incidents, or 25% were REARENDs where another MV REARENDed a MC.

    By the way, the most outstanding number when comparing MC crashes to overall MV crashes is the higher incidence of SINGLE_VEHICLE MC crashes — 40% versus only 17%.

    There is a small amount of impreciseness in these queries because the other traffic unit though usually another automobile or truck can be another MC, or a ped, or a bicycle.

    Simply plug in u.BodyStyleDesc LIKE 'MOTORCYCLE_MC%' in place of u.eUnitType IN ('PEDALCYCLIST') That replaces all the counting of bicyclists with motorcyclists. 
    
    SELECT eCollisionManner,count(1) FROM 2012_incident i WHERE EXISTS 
     (SELECT 1 FROM 2012_unit u WHERE u.IncidentID=i.IncidentID AND 
       u.BodyStyleDesc LIKE 'MOTORCYCLE_MC%' AND u.UnitNumber IN (1,2,3,4,5,6) ) 
      AND InjurySeverity IN (1,2,3,4,5)  GROUP BY 1 ORDER BY 1 ASC;
    SELECT 0 INTO @b; SELECT 0 INTO @a; SELECT a.eCollisionManner, @a, a.c, a.r, @b, b.c, b.r, a.r/b.r, b.r/a.r FROM (SELECT eCollisionManner, count(1) c, @a, count(1)/@a r FROM (SELECT *, @a:=1+@a FROM 2012_incident i WHERE NOT EXISTS (SELECT 1 FROM 2012_unit u WHERE i.IncidentID=u.IncidentID AND u.BodyStyleDesc LIKE 'MOTORCYCLE_MC%')) x GROUP BY 1) a LEFT OUTER JOIN (SELECT eCollisionManner, count(1) c, @b, count(1)/@b r FROM (SELECT *, @b:=1+@b FROM 2012_incident i WHERE EXISTS (SELECT 1 FROM 2012_unit u WHERE i.IncidentID=u.IncidentID AND u.BodyStyleDesc LIKE 'MOTORCYCLE_MC%')) y GROUP BY 1) b ON a.eCollisionManner=b.eCollisionManner GROUP BY a.eCollisionManner ORDER BY 3;
    
    SELECT eCollisionManner,count(1) FROM 2012_incident i WHERE EXISTS 
     (SELECT 1 FROM 2012_unit u WHERE u.IncidentID=i.IncidentID AND 
       u.BodyStyleDesc LIKE 'MOTORCYCLE_MC%' AND u.UnitNumber IN (1,2,3,4,5,6) ) 
      AND InjurySeverity IN (1,2,3,4,5)  GROUP BY 1 ORDER BY 1 ASC;
    
  5. One of the Human Factors / Naturalistic driving studies:

    Lee, S. E., Llaneras E., Klauer S., Sudweeks J. (2007) Analyses of Rear-End Crashes and Near-Crashes in the 100-Car Naturalistic Driving Study to Support Rear-Signaling Countermeasure Development, DOT HS 810 846

    Of the 7,024 observed rear-end events, 45 percent involved a decelerating lead vehicle, 38 percent involved a stopped lead vehicle, 2 percent involved a slower moving lead vehicle, and 15 percent occurred under various other situations. Crashes were predominately characterized by situations in which the lead vehicle was stopped, whereas near-crashes and incidents were more evenly distributed across instances of both stopped and decelerating lead vehicles. The majority of rear-end crash events (59%, or 16 out of 27) involved a stopped lead vehicle, while 22 percent (6 out of 27) occurred under conditions of a decelerating lead vehicle.

    i.e. few rear-end collisions involve “slow” vehicles.

    https://www.facebook.com/groups/427586350662325/permalink/929676697119952/

Comments are closed.