Poor youths who attended remedial-education classes while working part-time in the summer scored substantially higher on math and reading tests than a control group who held full-time summer jobs, according to a study released yesterday.

A Philadelphia nonprofit organization, Public/Private Ventures checked Boston; Fresno, Calif.; San Diego; Seattle, and Portland, Ore., in its study.

In each city, 150 youths of both sexes and varied races, ages 14 to 16 were enrolled, and a control group with similar background, educational attainment levels and age was established for the continuing experiment. The results reported yesterday covered findings for the last two summmers.

Members of the experimental group were required to participate for a full day, five days a week and for six to eight weeks and were paid the minimum wage for time at work and in remedial education.

However, they actually worked at federally funded summer youth jobs slightly less than half of the time. The rest was spent in remedial math and reading classes or in classes, field trips, lectures and discussions on ways to avoid pregnancy and on responsible social and sexual behavior.

The control groups worked full-time at summer youth jobs.

All participants and control-group members were tested at the start and end of each summer.

The organization said that the average student normally loses math and reading skills over the summer, then picks up in the fall.

The experiment found losses for all groups over the summer. But youths who entered the part-work, part-study group in 1986 showed a far smaller loss of skills than the control group, so that children in the experiment started school in the fall with a substantial advantage over the control group's members.

Thus, while the experimental group and the full-time workers scored about the same in reading at the start of last summer, by the end of the summer experimental-group reading scores were about six-tenths of a grade higher than those of the youths with full-time jobs.

Similarly, math scores for those in the remedial program were one-half academic year better by the end of the summer than the scores of the youths who worked full-time.

The experiment is funded by the Ford Foundation, William T. Grant Foundation, Aetna Life and Casualty Foundation, Ahmanson and Edna McConnell Clark foundations and similar groups, as well as the departments of labor and health and human services.