WASHINGTON Congressional auditors say the Federal Motor Carrier Safety Administration has improved the quality of data in its SafeStat program but should alter the method it uses for calculating bus and truck fleet safety ratings.

SafeStat, which is used by FMCSA to determine which bus and truck operators to target for safety enforcement actions, has been the focus of considerable criticism for years. (See March 1, 2004, Bus & Motorcoach News.)

Both government investigators and commercial vehicle operators have challenged the quality of the programs data, as well as the way it generates

safety rankings.

In a reported issued last month, the Government Accountability Office said the FMCSA could improve its ability to identify carriers most at risk of having crashes by 9 percent if it would use a 'binomial regression model' to analyze raw safety data. Binomial regression modeling is a statistical procedure that can be used for the analysis of data.

'While SafeStat does a better job of identifying motor carriers that pose high crash risks than does a random selection, regression models GAO applied do an even better job,' the report said.

SafeStat currently includes four pieces of data: driver information, vehicle safety information, safety management information and a companys crash record.

FMCSA adds the four pieces together to produce a score, with the crash piece weighted the heaviest. The GAO and other critics note that the formula never changes regardless of whether past crashes are the best predictor of future ones.

'But in the regression model, its whatever fits best with the accident,' said Jim Ratzenberger, GAO assistant director. 'It allows for change over time.'

Using the new statistical formula would mean 'more effective targeting of compliance reviews to the set of carriers that pose the greatest crash risks,' GAO said.

The agency said their statistical model 'experienced 9,500 more crashes than those identified by the SafeStat model over an 18-month follow-up period.'

That result led the congressional watchdog agency to recommend FMCSA 'apply a negative binomial regression model, such as the one discussed in this report, to enhance the current SafeStat methodology.'

GAO also said it expected 'shortly' to issue a study on how FMCSA identifies and takes action against carriers that are 'egregious safety violators' and 'how thoroughly and consistently FMCSA conducts compliance reviews.'

While urging the FMCSA to improve its scoring system, GAO did acknowledge the federal agency had made strides.

According to GAO, the median number of days it took for states to report crashes dropped from 225 in 2001 to 57 in 2005, and the number of crashes reported within 90 days rose to 89 percent in fiscal 2006, from 32 percent in fiscal 2000.

The U.S. Department of Transportation inspector general office also is conducting a review of SafeStats data accuracy. After a 2004 report by the USDOT inspector general, FMCSA removed the scores generated by SafeStat from its Web site (See June 15, 2004 Bus & Motorcoach News).

In 2005, GAO said the system for collecting data had improved, but the datas quality still needed strengthening.

Last year, FMCSA Administrator John Hill said he supported putting the safety rankings back up for public viewing, but would wait until the inspector generals review. An FMCSA spokeswoman says Hill 'still wants to wait for the Inspector General report before making a decision on reposting the SafeStat scores.'