FOURTEEN YEARS ago, some friends and I abandoned our various careers to found a school for young male delinquents on a remote island off the Massachusetts coast. We were part of a program to replace the state's much-criticized reform schools and our idea was to put these boys in a self-sufficient community where honesty and cooperation would contribute so visibly to everyone's well-being that our students would adopt these values as their own.

We were confident. Taking our cue from the social scientists, we assumed that a delinquent's behavior was entirely logical, given his background: He stole because society did not let him work, and he turned to violence because violence was the only way to survive in an urban jungle. We gave him credit for making rational choices based on very limited options. We believed that by expanding those options we could lead him to the "right" decisions.

On the other hand, what we saw in the teen-agers who came to Penikese Island was what seemed to be completely random behavior.

What kind of logic, we wondered, would lead a boy to weep at the memory of his brother who had died from a drug overdose and then go and overdose himself? We puzzled over why our kids were as unmotivated by the carrot -- learn to read so you can get a better job -- as they were undeterred by the stick -- steal that car and you'll go to jail.

But so firmly rooted was our thinking in the social scientists' quest for causal relationships that it wasn't until after several years of beating our heads against the wall that we finally realized the truth: What appears to be random behavior is random behavior.

The problem was not that the delinquents we work with have been badly programmed by their environment. They haven't been programmed at all. In contrast to the child who grows up with the security of knowing that any action on his part will lead to a predictable reaction on the part of those who care about him, our kids live in a nightmare world of conflicting signals.

What, for instance, is Dan to think when his father lectures him piously about the evils of crime one minute and then boasts of his own criminal exploits the next? How many cars should Fred conclude it is okay to steal when his caseworker takes him to task for "excessive car stealing"? Do we see hope for Jim when he advises us that "If my kid brother screws up like me I'll break his expletive arms"? Is Ron likely to stop beating on people if his father boasts with ill-concealed pride that "even the gym teacher's afraid of my kid"?

For boys like these four, what they see contradicts what they hear. What they do has no identifiable influence on what happens to them. Whether they are praised, punished or ignored depends more on the erratic moods of unstable adults or the often equally mercurial intervention of officialdom than on anything they do themselves.

In an unpredictable world, their only guide to behavior is simply to satisfy the impulse of the moment. A childhood that never gave them the chance to make consistent connections between cause and effect leaves them feeling powerless to influence the course of their lives.

This sense of helplessness leads both to the kind of unfocused anger that explodes easily into violence and the capacity for impersonal cruelty that is characteristic of those who cannot relate the effects of their own behavior to themselves, much less to others. Mental-health programs get the boys whose reaction to powerlessness is suicidal withdrawal; we get the ones who lash out more violently against their sense of impotence.

The failure to recognize this maddening randomness is why so many of the programs that briefly catch the public eye with dramatic claims of success end up succumbing to frustration.

A decade ago, when the crusade to "deinstitutionalize" delinquents and mental patients was at its peak, these claims were being made by the idealistic advocates of a whole range of noncoercive behavior-modifying therapies. These reformers promised to improve on what was alleged to be the dismal record of their more "punishment-oriented" predecessors.

In the years since, the pendulum has begun to swing once again away from bravely naive assumptions that an environment of unstructured and non-judgmental "caring" was enough to dissuade kids from crime. Today's more hard-headed realism is reflected in programs that square away young punks in pseudo boot camps, scare them straight through jailhouse visits with violent criminals, or inspire them to rectitude during arduous wilderness odysseys that lend themselves nicely to enthusiastic media coverage.

The Penikese Island School began by attracting the kind of publicity that quixotic pursuits of this kind always get, particularly if they occupy picturesque locations. The coverage we received contained all the predictable superlatives. Our "no-nonsense" curriculum was "unique." Our staff was "dedicated" and our students were quoted talking earnestly of "turning themselves around."

All of this was true.

Yet, when five years later we did a survey involving interviews with as many as we could locate of our first 100 students and surveys of Department of Probation records on the entire group, we discovered what, by then, we already had begun to suspect. All but 16 had not turned themselves around; 84 had gone on to lives destructive in varying degrees to themselves and society.

Among those 84, most were survivors who had mastered the chameleon-like ability to adapt instinctively to whatever an unpredictable fate threw at them. This is why many of them did so well at Penikese.

But the same profound lack of any sense of self that allowed them to conform so readily to our expectations while they were with us made them equally malleable to the evil influences they reencountered when they returned to the environment that produced them.

This is why I am so skeptical of the enthusiastic claims made by so many of the programs I have read about. On the face of it, Penikese would seem to be accomplishing all of the same things our more optimistic colleagues say they are doing. Our students do seem to adopt the values we encourage on the island. They demonstrate the kind of work habits that could make them competitive on the job market, and they learn enough carpentry, cooking, farming and boat handling to enter any of these fields on an apprentice level.

But if our follow-up study is even close to representative, it appears that only a small minority acquire from this experience a firm enough set of values to steer their own course. In the long term, most continue rudderless, swept along by the strongest currents they encounter.

There may well be others who do better than we do in this business, but somehow I doubt if any who work with the same kids that we do can do four times better, as some of the published figures would indicate.

This is not, however, to accuse anyone of conscious fraud. There is a lot of wishful thinking in the human services. The people who commit themselves for piddling salaries to helping others have a big emotional stake in the success of their charges. They watch kids come to them sullen, scared and pale from too long in detention, and a few months or even weeks later, they see the same kids graduate, now tanned and healthy and filled with the confidence of having finally succeeded at something. Faced with so dramatic a transformation, claiming a success is understandable.

Sadly, most of these claims will not stand up. Next time you come upon a euphoric description of some group succeeding at "saving kids," look to see how precisely the criterion for success is defined. If the measure used is the number of graduates who do not reappear in juvenile court, then one must deduct from the percentage given all those who, because of age, have moved on to the adult criminal justice system.

Beware also the negative test. Simply not showing up again in court does not by itself constitute success. In our study, the 16 percent were those who had freed themselves entirely from the corrections, mental health, and welfare systems and who had been productively occupied for a year or more. Look at how long the program described has been in existence. Often those trumpeting success simply have not been around long enough to compile meaningful figures.

Inquire how the data used was collected. Kids in trouble lead such nomadic lives that it is not always easy to find out what happens to them. One of our staff, who knew and was trusted by most of the kids she was looking for, spent several months crisscrossing the state trying to find the graduates we surveyed.

If this kind of scrutiny is applied to many of the figures given, I suspect we will find that what they are showing is a remission of symptoms rather than a cure. Many of our graduates do look like successes for however long our influence can sustain them after they get back home. Some return to school. Others find jobs with civic-minded employers. For a couple of months all seems to be going well, and then they lose it.

Ralph is a case in point. After he left Penikese he got a job cleaning up a movie theater. Six months later he was promoted to manager. We were proudly using him as an example to our other kids when I got a letter: "Hey George, I got good news and I got bad news! Good news is, I'm gonna be a father. Bad news is, I'm in Walpole prison ." He had, as he put it "got in with the wrong gang."

Two questions arise from all this. The first is: Why are we doing so badly?

Heredity, nutrition, poverty, drugs, teen pregnancies, the profoundly confusing American ethic that glorifies violence when it is used on the side of right; all these things play a part. But the one certainty is that their combined effect is producing a lot of young people whose problems are beyound the abilities of any short-term program to help, if in fact they can be helped at all. No experience of a few months, no matter how positive, can undo the damage of the preceding 17 years.

So the next question is: Is it worth it? Of course it is. First of all, that 16 percent, if the figure is accurate, is not a bad return for our efforts, particularly if we consider what those few successes would cost themselves and society in misery and dollars if they hadn't "made it." Certainly, no hospital would make less effort on behalf of a patient having only a 16 percent chance of recovery, any more than we should allow a realistic assessment of what we can accomplish to discourage us from trying to help kids. Every boy who comes to us is a potential member of that 16 percent.

The other thing to remember is that statistics provide an incomplete picture of actual results. If we are all the sum total of our experiences, both good and bad, then giving a kid who has had only bad experiences one good one is not wasted time. We may, in the process, reduce his capacity for crime from murder to car theft -- and that is progress, even if it will never show up in the statistics.

The danger done by inflated claims of success is that they make it easier for us to ignore the really fundamental changes our society must make if it is to stop producing the subculture of uncentered young people who are passing unaffected through our rehabilitation systems. Sinking more money into correctional programs will do less, for example, to reduce juvenile violence than would eliminating the media-created role models that so powerfully influence troubled young people to measure their own effectiveness in terms of their ability to hurt people.

To stop delinquency we must do something about its cause. At Penikese we have tried to create a small part of the world where everything makes sense to the boys who live there. We hope that from such simple lessons as learning that they must cut wood to keep themselves warm will come the realization that they can control their own lives for better or worse by the things they do. But as long as kids continue growing up in a larger world that is so profoundly confusing, this lesson will be too little and too late for most of them.

George Cadwalader is founder of the Penikese Island School.