By the age of 14, 1 of every 10 girls across the nation and almost 3 of every 10 boys have had sexual intercourse, we are told in "Sex and the American Teenager." By age 18, the figure is 53 percent for girls and 46 percent for boys.

This book is filled with assertions like these, based largely on a survey of 1,067 American teen-agers that was conducted for Rolling Stone magazine a number of years ago. It is to be the subject of an HBO broadcast June 10. The nation's young, apparently, are doing more about sex than just listening or dancing to rock songs with explicit lyrics.

There is a problem, however, with the accuracy of these and many other conclusions offered. The survey suffers from a low and possibly biased response rate among whites, and too few interviews among nonwhites.

The division of labor for the book was such that Geoffrey Stokes, a New York writer, did the introduction and the chapters relating the survey findings. Coauthor Robert Coles, the child psychiatrist and teacher, contributed a chapter at the outset describing how young people he has encountered experience sexuality (often through fantasy), along with a chapter of commentary at the end.

The result of the Rolling Stone survey, as Stokes puts it, is a portrayal of a generation that is partly conservative, partly liberal and in other ways "as confused as any generation that has preceded them." That is probably an apt description; why should they be otherwise? But it is more an educated guess than a conclusion based on reliable data.

Many people, adolescents included, will not discuss their sexual practices and opinions. The subject is just too sensitive. The national firm that did the survey -- and which would not allow its name in the book -- started with a sample of 3,000 households but got only a 34.2 percent response rate, or 1,026 interviews.

Both Stokes and Coles discuss the question of the survey's validity, and they end up on different sides. Stokes notes that while the response rate was not what had been hoped for, other surveys on sex have done considerably worse, and he concludes that the Rolling Stone work provides "the most accurate picture yet developed of sex and teenage Americans." Coles, however, questions the project's reliability on several grounds. "One has to believe," he writes, "that the more conservative part of the teenage population may be less represented here because of a reluctance to participate."

For every youngster who took part, two declined. If those two are more conservative sexually, as Coles suggests, then "Sex and the American Teenager" could be sharply exaggerating the sexual activity of the nation's young. We have no real way of knowing. Coles also casts doubt on the responses of adolescents who did participate. Again and again, he writes, young people have told him that they respond to surveys with answers they think they should be providing, rather than honest ones.

Unlike a voter poll, which may be validated or proven wrong by election results, there is no easy check on this Rolling Stone survey. It may well be that Stokes and Coles are both correct. The survey could be the "most accurate" yet done on teen-age sex and still be wide of the mark.

Thus it is not only disheartening but frustrating as well to be told that 1 of every 6 teen-age girls becomes pregnant, that the pregnancies almost always end in abortion (86 percent of them) or that 14 percent of all teen-age girls have been raped, one quarter of the time before they are 11 years old. Disheartening because we all know that teen-age pregnancy, abortion and rape are serious national problems; frustrating because we have no idea whether "Sex and the American Teenager" is anywhere near being precise in its attempt to quantify these problems.

Also included in the book are 48 pages of verbatim excerpts from conversations between interviewers and the adolescent respondents, in order, as Stokes puts it, "to give the teens a chance to speak in a context where their words would be free from our judgments."

You've heard the expression "raw data" used in reference to polls and surveys? Well, these are raw conversations, often extremely raw. They tell us nothing about teen-agers as a group, but they do demonstrate that some youngsters lack knowledge of sex, that some are pretty experienced and that many can speak as foully as any adult.

I saw no mention in the book of when the Rolling Stone survey was conducted, information usually considered relevant in the reporting of such work. A Harper & Row editor told me it was done in 1981 or 1982. A researcher cited as having put the questionnaire together, however, told me it was conducted in 1979 or early 1980. Either way, that dates the material considerably. Since then, it is likely that the problems of AIDS and genital herpes have profoundly changed many people's sexual behavior, and it may well be that a profile of today's teen-agers would yield different results in many regards. That wouldn't make it any more reliable, just more up to date.