One morning in February 1991, Sacramento, Calif., residents woke to a thunderous blast when a fully loaded gasoline-tanker truck rolled and skidded 100 feet and landed on a parked car at the edge of a tranquil suburban neighborhood. On July 14, 1991, a Southern Pacific freight train filled with toxic chemicals derailed at a bridge near the small town of Dunsmuir, Calif., falling 40 feet into the Sacramento River below.

To many, these might seem to be two incidents with no connection. One damaged property for blocks and burned two homes to the ground. The other spilled tons of weed killer into the Sacra- mento River, killing virtually every organism along 40 miles of river. Michael Parrish of the Los Angeles Times sensed something might be wrong with the hazardous material transportation in this country when he saw the two disasters.

Sparked by the thought, he began researching the transportation of hazardous materials, or 'hazmat,' by both rail and truck, to put together a picture of the national scene. Parrish's speculation proved to be on target when he obtained computer tapes covering the past ten years from the U.S. Department of Transportation's Hazardous Material Incident Reporting System. A total of 67,657 spills had been reported to the Department of Transportation from January 1982, through December 1991.

'There is a story here,' Parrish thought. Parrish needed someone to transform the data on those computer tapes into comprehensible information. He enlisted the help of Richard O'Reilly, the Times' director of computer analysis. For the past three years, O'Reilly has devoted his time to doing data projects for the reporting staff. O'Reilly took the tapes from Parrish and used an IBM mainframe computer and the data analysis software SAS to conduct the research.

Unlike database programs which use structured query language to sort data, SAS needs a tailor- made program to analyze the data. Each time O'Reilly analyzed a new set of data, he needed to write a new program. But SAS helped him do hundreds of cross-tabs and frequency counts on which areas of the country had the biggest problems with hazardous material and whether particular kinds of chemicals had problems with particular kinds of containers.

A number of significant findings emerged. He found that from 1982 to 1991, hazmat incidents rose 37 percent. Incidents involving trucks, which carry most of the hazmat, went up 34 percent. Injuries to people as a direct result of the truck spills soared 374 percent. Incidents on the nation's railroads were up 36 percent. And almost all the deaths, 106 out of 108, involved tanker trucks. O'Reilly had talked about the structure of the database with people at the Transportation Department earlier to avoid pitfalls.

Still, he encountered problems. There were three sets of data within the database: the incident report, the containers report and the remarks report. In a train accident, each car of the train would be counted as separate spill, but there was actually only one train wreck. In the container database, for example, if there was a glass jar inside a cardboard box and the box was one of four within a larger carton, there would be a separate report on the same spill for each of the containers.

Each layer the material leaked through was counted as a different spill. To solve this problem, O'Reilly spent hundreds of hours producing a summary report of all the different findings and sent it to the Department of Transportation for spot-checks to verify his methodology and findings. The people there got the same answers as O'Reilly. With the data in hand, reporters Parrish and Tom Furlong interviewed police officers, government officials, highway accident investigators, critics, hazmat specialists, engineers, rail inspectors and package delivery workers.

The result was 'Danger in Transit,' a three-day series in the Times. Appreciation letters and phone calls started coming in soon after the series appeared Sept. 20-22, 1992. Readers told the Times that they didn't realize the danger of the hazmat transportation until reading those articles. 'That (computer-assisted reporting) was terrific research,' said Furlong, whose article in this series was seen on Sept.

21. 'Using that information, we were able to go out and document the accidents and explore what had happened That may be the best journalism you can do.'