BOOKS based on survey research are often

dreary. Ideas tend to become lost in discussions of methodology and computer tests of significance. This is not a new problem. Over the years, chi squares and other statistical measures have become their own reward, and pity the poor reader.

So on the one hand, it may be that congratulations are in order for Philip Blumstein and Pepper Schwartz, two sociologists at the University of Washington. Though their work is largely the result of interpreting thousands of questionnaires, their chief interest is in giving a lively account on a subject of enormous importance. No one can accuse them of being too concerned about methodology.

American Couples describes how some married people, some unmarried men and women, some homosexual men and some lesbians succeed or fail in their drive to spend much of their lives with just one other person. If one aim of sociology is to help people examine the way they live, then the book may be considered a success. It does that quite well. The cast of characters is so diverse and the problems so varied that almost every reader will be able to relate portions to his or her own life. That is the book's chief merit. It encourages introspection.

On the other hand, American Couples has major, insurmountable flaws. It is likely to be an embarrassment to the National Science Foundation, which gave the authors a grant of $236,000 to do the research.

Blumstein and Schwartz say they sent 38- page questionnaires to 7,397 heterosexual couples, 1,875 male homosexual couples, and 1,723 lesbian couples, all of whom had volunteered to take part. These participants had heard of the project through advertisements, other publicity, or word of mouth. A little over half (55 percent) returned usable questionnaires.

The sample therefore was self-selecting, one of the principal no-nos in any survey research that attempts to make generalizations about the public at large. People who volunteer to take part in surveys about sensitive aspects of their lives are likely to be quite different from other, less forthcoming people.

In addition, all or almost all of the research took place in three locations: New York, San Francisco and Seattle. Those are nice cities, but few would consider them representative of the nation at large.

The authors themselves note that "the way in which participants were sought did not guarantee a sample representative of the population of the United States." That is putting it mildly. There is simply no way that conclusions drawn from these interviews can be applied to any people other than the ones interviewed. The authors hedge the problem, and never say that in so many words.

Instead, set off in large, italicized type every few pages are conclusions like these:

"When partners are disappointed with the amount of money the couple has, they find their entire relationship less satisfying, except among lesbians."

"When a wife works, the couple fights more about how the children are being raised."

"Women want more time to themselves than men do."

"Heterosexual men who receive oral sex are happier with their sex lives and with their relationships in general. Men who perform oral sex are also happier."

Unfortunately (or, perhaps, fortunately), there is no way of knowing whether any of these statements is valid. Nevertheless, in advance of its publication, American Couples has received extensive publicity on network television and in news magazines and major newspapers, almost all of it a recounting of the book's "findings."

Blumstein and Schwartz say that "the issue of representativeness becomes less troublesome as the number of participants increases." That is absurd; it is like a merchant saying he can turn a profit on a loss-leader product if only he sells enough of it.

At the National Science Foundation, which is not in the business of issuing grants that turn into advances for trade books, program director Murray Aborn says the agency has long been concerned with the Couples project. The authors made no mention of plans for a popular book in their request for funding, and NSF expected their work to result in a monograph or other scholarly publication, Aborn told me.

The grant covered the period from 1977 through 1979. Aborn said NSF understood that the findings would not be representative of couples nationally. But, after considerable review, he said, the grant was issued in hopes that the study would gather insight into an area of obvious national concern.

Since then, Aborn said, "there has been considerable back and forth about the turn of events. . . . I am dismayed. Some of the things we were worried about way back and attempted to control happened nevertheless."

The book has other, non-technical weaknesses as well. In their extensive use of quotes, the authors seem to have selected especially vile ones from women. In addition, about a dozen pages are devoted to discussions of fellatio and cunnilingus, but there is only passing mention of AIDS and genital herpes, two very serious problems for many of the people who might look to this book for useful information.

In its behalf, American Couples offers a useful historical perspective that includes good bits of information, noting, for example, that "in 1960, 28 percent of American women between the ages of 20 and 24 had not yet married. By 1979, the figure had jumped to 49 percent." That is a very striking change in our society.

The authors let their "volunteers" do a great deal of the talking, especially in a section of the book that recounts personal interviews with 20 couples. Some of that talk seems cheaply sensational, but some, again, accomplishes one of the worthy aims of any book: to move the reader toward serious reflection.

In a way, these occasional worthwhile aspects make the book very exasperating. Without them, it would be very easy to dismiss American Couples as unsound and nothing more. Unsound it is, but it is also something more.