An Examination of “Fault,” “Unsafe Driving Acts” and “Total Harm” in Car-Truck Collisions Forrest Council (HSRC) David Harkey (HSRC) Daniel Nabors (BMI) Asad Khattak (UNC) Yusuf Mohamedshah (Lendis)

Study Goals Further exploration of crashes involving one large truck and one car –Examination of “fault” in full-distribution of car/truck crashes (rather than just fatals) –Crash-based verification of “Unsafe Driving Acts (UDAs)” based on past data and expert judgment –ID “critical combinations” of roadway facilities and locations based on “total harm” (i.e., frequency + severity

Past Findings Truck/car crashes are severe, with non-truck occupants much more likely to be the one killed or injured “High profile,” with much current emphasis on reducing truck-related fatalities (e.g., by 50% by 2009) Car drivers are much more likely “at fault” in fatal car/truck crashes –Contributor in 81% and only contributor in 70% “Fault” is less clear in studies of total car/truck crashes

Past Findings (cont) Based on expert-panel, top 12 of 26 UDAs for cars are: –Driving inattentively –Merging improperly, causing a truck to maneuver or brake quickly. –Failure to stop for a stop sign or light –Failure to slow down in a construction zone. –Unsafe speed (e.g., approaching from rear too fast) –Following too closely. –Failure to slow down for poor environmental conditions –Changing lanes abruptly in front of a truck. –Driving in the A no –3-way tie: Unsafe turning, primarily turning with insufficient headway Unsafe passing, primarily passing with insufficient headway Pulling into traffic from roadside in front of truck without accelerating sufficiently.

Past Findings (cont) A second FARS-based UDA study says: –Car/truck crash-maneuvers are not much different from car/car maneuvers Only 4 of 94 pre-crash maneuvers were more likely in car/truck crashes –Following improperly –Driving while drowsy or fatigued –Improper lane changing –Driving with vision obscured by rain, snow, fog, or dust. But these were found in only 5% of fatal car/truck crashes

Summary of Gaps While cars are more “at fault” in fatal crashes, little data on non-fatal crashes UDAs based on expert judgment need more crash-based validation No study associating critical crash types or maneuvers with specific roadway characteristics (for treatment ID)

Databases Used “Fault” analysis for total car/truck crashes – NC data, where a contributing factor was assigned in 97% of the cases (as opposed to 1/3 of cases in GES files) UDA verification –1999 GES data Roadway characteristics – NC crash and roadway inventory data

Study Methods – “Fault” Analysis “Fault” assigned to car or truck drivers if any contributing factor assigned by police Analysis of car/truck cases with “fault” for –Non-truck only –Truck only –Both –Neither Compare to earlier fatal-crash results

Results – “Fault” Analysis Overall, trucks more likely to be “contributing” than cars – 48% vs. 39% Trucks more “at fault” in following crashes –Backing, rear-end crashes, right-turn (non- crossing), left-turn (non-crossing), sideswipe Cars more “at fault” in –Head-on, angle, right-turn crossing Differs from fatal-only findings somewhat

Study Methods – UDA Verification Could match 13 of 26 UDAs from expert panel with subsets of GES crash data Ranked separately on frequency and severity of crash Compared combined rank to “Top 12” from expert panel

Results – UDA Analyses Could only define GES crash subsets to “match” 13 of 26 UDAs (6 of “Top 12”) Could not define #1 – “inattention” For those matched, subsets were only small part of total car/truck crash problem –For three of “top 12” – 2% to 5% of problem –For other three of “top 12” – less than 1% of problem Severity for these six was higher than for average car/truck crash. (Panel influenced more by severity?)

Results – UDAs (cont) 13 then ranked based on both frequency and severity Combined ranking was sum of two ranks Some general agreement with past rankings –Unsafe speed, failure to slow for poor weather, changing lanes abruptly But some significant disagreements –Driving left of center #1 here, but not in “top 12” –Failure to slow in construction zone #4 by panel but #13 (last) in these rankings

Study Methods – Critical Roadway Combinations Assigned a “harm cost” to each car and truck based on severity level (fatal, injury, PDO) and thus to each crash –“ Harm cost” based on comprehensive cost guidance from NHTSA and OST Categorized crashes by 11 facility types (e.g., rural interstate), 7 crash types (e.g., head-on) and 6 location types (e.g., driveway, signalized intersection) Calculated “average harm cost” for each combination using regression model (to smooth the data) Calculated “total harm cost” by multiplying by frequency

Results – Critical Combinations 325 NC combinations had sufficient data for “total harm” calculation Top combination -- angle crashes at stop/yield intersections on undivided rural minor arterials and major collectors Second – “angle” crashes on urban interstates (e.g., collision during passing or merging) “Top” facility types – undivided rural minor arterials and major collectors, undivided rural principal arterials (i.e., interstates lower) Top group included some high severity/low frequency (e.g., head-on) and some low-severity/high-frequency (e.g., rear-ends), supporting use of “total harm” analysis

Summary Cars are clearly not as “at fault” in total crashes, meaning that trucks should be targeted for intervention also UDAs rankings by expert panels do not agree well with rankings based on crashes. If used for treatment development and targeting, better methods than expert rankings are needed. “Total harm” can successfully be used to identify critical crash and roadway combinations, and would appear to be a feasible treatment targeting method.