For years the auto insurance industry has given large insurance discounts for children who take drivers' education courses, because statistics show that they have fewer accidents.

The preliminary results of a new major study, however, indicate that driver's education does not prevent or reduce the incidence of traffic accidents at all.

Many researchers conclude that drivers' education students have lower accident rates not because of their training but because the sort of children who volunteer for it have different social backgrounds and upbringing from those who don't. One study, for example, shows that students who get high grades have fewer accidents than those who don't.

In a study of about 17,500 children in DeKalb County (Atlanta), Ga., the National Highway Traffic Safety Administration finds no difference in the number of accidents and traffic violations between children who take the standard high school drivers' education course, or its equivalent, and those who have no drivers' education at all.

These results back controversial studies that show drivers' education not only does not decrease accidents among 16 to 18 year olds, but actually many cause thousands of additional traffic deaths every year, according to Leon Robertson, a Yale University researcher.

He did a study of 10 Connecticut towns that eliminated drivers' education from their schools which showed that "about 75 percent of the 16-17 year olds who could be expected to have been licensed if they had taken high scool driver education waited until they were 18 or older to be licensed when high school training was no longer available."

The result of having fewer licensed children on the road, according to Robertson, was a large decrease in the number of serious accidents in those Connecticut towns.

The Robertson study has begun to have some effect on drivers' education around the country, and at least one community -- Farmington, Conn. -- has eliminated the course largely on the basis of Robertson's work, according to Farmington school superintendent William Streich.

"We were encouraging [teen-agers] to drive by offering the course in high schools. By not offering it, we may discourage it and postpone licensure," Streich said.

In the $4.2 million NHTSA study, students who volunteered for drivers' education were divided into three groups: those who would get no formal drivers' education, those who would get a special, intensive 72-unit course including training on a special driving track.

The director of the NHTSA study, Clay Hall, said it showed no statistically significant difference" in accident figures among all the groups after three years of the study's six-year run. He said he expected the preliminary numbers to be born out in the rest of the study, but that the interim report would draw no final conclusions.

Robertson's study has been attacked by a number of groups, including the professional association of drivers' education teachers, as unsound statistically.

Ray Burneson, traffic safety specialist with the National Safety Council, criticized the study, saying that it was a product of a group (NHTSA) that was run by people who believe "that you can't do anything to train drivers. You can only improve medical facilities and build stronger cars for when the accidents happen. . . . This knocks the whole philosophy of education."

Hall said the study would note the "favorable trend" that children in the special course are getting 16 percent fewer traffic tickets than those who haven't taken a drivers' education course. He said this shows that drivers' education can have a positive effect on accident rates.

The training part of the program is now finished and the period in which the records of the children are followed in regular highway driving has begun. The interim report is due out next week and the final report is planned for 1983.