Study: Crash Data Flawed

This story appears in the Jan. 26 print edition of Transport Topics.

Safety regulators do not have a foolproof path to follow for quickly assessing fault in crashes and then using those judgments to predict future risk among trucking companies, according to a study released last week by the Federal Motor Carrier Safety Administration.

The study examined nearly 11,000 police reports on accidents involving trucks and buses, and found discrepancies among the reports and no clear way to use fault assessment to produce a superior model for predicting future crashes as part of the Compliance, Safety, Accountability program.

Produced by MaineWay Services, the study tended to confirm FMCSA’s current statistical usage but also said that, if the agency were to use more police reports in finding fault, it would cost between $3.9 million and $11.2 million annually.

FMCSA has been using simple accident involvement as a factor in the calculations that are the underpinning of the CSA program. American Trucking Associations has argued that merely being in an accident is not a useful tool in predicting the likelihood of future crashes. Instead, the truckers have said, safety regulators should look at poor operations by carriers and drivers that lead to avoidable accidents.

FMCSA has argued to the contrary and cited work by university and other researchers showing the relationship between one crash and a future crash.

As for altering the model based on the carrier’s role in the accident, the Jan. 21 report said the change “does not appear to improve its ability to predict future crash rates when all crashes are considered.”

“Numerous times over the past five years, ATA has respectfully requested FMCSA to screen out crashes from CSA where it is plainly evident the professional truck driver and motor carrier were not at fault,” said Dave Osiecki, the federation’s chief of national advocacy.

“Instances where a truck is rear-ended by a drunk driver, or hit head-on by a motorist traveling in the wrong direction on the interstate, or as happened just [Jan. 19] when a truck was struck by a collapsing bridge are clearly not the fault of the professional driver and certainly should not be used to target his or her carrier for potentially intrusive government oversight,” Osiecki said.

FMCSA is accepting comments through Feb. 22. The agency was asked what would happen after the comment period but offered no response by press time for this issue.

The MaineWay study started by looking at 10,892 police accident reports. Of that total, 88.3% of the reports were found to meet the study’s criteria for analysis and have sufficient information.

Problems with the police reports arose when compared with the Fatality Analysis Reporting System database kept by the National Highway Traffic Safety Administration.

There were strong matches between FARS and the police reports on weather conditions and roadway surface conditions, 95.7% and 96.7%, respectively. In three other areas, though, agreement disintegrated markedly.

Agreement on traffic flow was only 52.4% between the two types of reports and, for designating a first harmful event, there was a match 46.9% of the time.

For listing driver-contributing factors, the FARS-police report match rate was just 12.6% of the time. There were discrepancies in 5.3% of cases and missing police report data 82% of the time.

“The agency was unable, in this type of analysis, to establish which record, the PAR or FARS, was more accurate but simply identified the fact that the two data sources were not in agreement,” the consultant’s study said, referring to the police accident report.

The peer-reviewed paper included comments, such as those of Eric Teoh, a senior statistician for the Insurance Institute of Highway Safety. Teoh said the appeal of examining crashes where blame is assigned to the truck driver or carrier is that such crashes “should be a good predictor of future assigned crashes.”

In contrast, when another party is at fault, Teoh said, such accidents are “largely due to exposure to other vehicles or adverse environmental conditions.”

Teoh also said large trucking companies were at a disadvantage.

“As the number of power units of a carrier increases, all else being equal, so does the likelihood of having a crash of any type,” he said, and he suggested looking at accidents on a per-truck basis.

Attorney Rob Moseley was critical of FMCSA’s statistical practices, saying the agency should separate at-fault from no-fault accidents.

“FMCSA has never made the distinction between the two. If they backed out the at-fault accidents and still found a significant correlation between not-at-fault and future accident, then we can talk,” said Moseley, the head of the transportation group at Smith Moore Leatherwood.