A new research report finds that while some CSA scores are accurate predictors of crash risk, others work in the opposite direction.

The report by the American Transportation Research Institute, an arm of American Trucking Associations, uses statistical analysis to see how scores in the five public BASICs relate to actual crash involvement.

It found a strong correlation between score and risk in three Behavior Analysis Safety Improvement Categories. Scores in the Unsafe Driving, Fatigued Driving and Vehicle Maintenance BASICs are positively related to crashes, the report says.

But there is a negative relationship between score and risk in the Driver Fitness and Controlled Substances and Alcohol BASICs, the report says. In fact, the higher the score the lower the risk.

"ATRI's research identifies a key weakness in FMCSA's Safety Measurement System," said Scott Mugno, vice president of safety for FedEx Ground, in a statement.

"The conclusions in ATRI's study support what many motor carriers have found to be true in their operations - namely, that scores in the CSA Driver Fitness BASIC do not bear a statistical correlation to crash risk," he said.

"However, the industry has always supported CSA where it does reduce crash risk and ATRI's study validates that there are portions of CSA that are working as intended."

The Unsafe Driving BASIC provides the strongest indicator of risk. A carrier with a score of 99 is about three times more likely, on average, to be involved in a crash compared to a carrier with a score of zero, the report says.

On the other hand, as the Driver Fitness score rises, the relative crash risk declines.

This led the ATRI researchers to conclude that the Federal Motor Carrier Safety Administration's method for calculating the percentile scores is incorrect.

"It is likely that FMCSA's severity weighting methodology places too much weight on safety-irrelevant violations and too little weight on safety-critical violations in (the Driver Fitness and Controlled Substances and Alcohol) BASICs."

Based on these findings, the Institute suggested that the agency consider changing the way it relays CSA information to the public.

Rather than using the percentile scoring system, the agency should show the carrier's relative safety status by placing it in one of three categories based on increasing risk.

The first category would be for carriers that have enough data in at least one BASIC, but no score. The second would be for carriers with scores in at least one BASIC, but no "Alerts" reflecting severe violations.

The third category would be broken down into five sub-groups, based on the number of "Alerts" a carrier has.

The agency could continue to use percentile scores for internal purposes, but this approach to public data would address concerns about shippers and others taking the wrong message about a carrier's safety risk from the BASIC data now on the agency's website, the report says.

Related Stories:

9/14/2012 House Committee Hears CSA Complaints

9/4/2012 Group Effort Needed to Deal with Crash Accountability Question