LATE ONE gloomy winter afternoon in 1980, New York sociologist Robert Martinson hurled himself through a ninth-floor window of his Manhattan apartment while his teen-age son looked on. Martinson had become the leading debunker of the idea that society could "rehabilitate" criminals. His melancholy suicide was to be a metaphor for what would follow in American corrections. The question that obsessed Martinson still haunts debate on the purposes of corrections. It surfaced recently in the controversy over Maryland's Patuxent Institution; and it underlies drug czar William Bennett's plan for more prisons and Mayor Marion Barry's call for 2,000 more cells. They embody the now-popular assumption that rehabilitation is pointless. Policy initiatives emphasize more cells for more inmates, and the main federal "message" to cities, as Attorney General Richard Thornburgh succinctly put it, is: "Do the crime, you do the time." But the present fixation on punishment and deterrence may prove a costly mistake. In fact, there is considerable evidence that rehabilitation which adheres to certain principles can be dramatically successful {see box.} And we'd better start learning what those standards are. The currently fashionable notion of more "hard time" for more offenders could bankrupt many state and local budgets while guaranteeing even higher recidivism rates. Martinson's skepticism about rehabilitation derived from his role in co-authoring a 1975 survey of 231 studies on offender rehabilitation spanning the previous 30 years. Titled "The Effectiveness of Correctional Treatment," it became the most politically influential criminological study of the past half century. The time was ripe: From 1963 to 1973, murder, assault and burglary rates doubled while robberies tripled. Martinson's views were enthusiastically embraced by the national media, often under the headline, "Nothing Works!" Yet curiously, all the sturm und drang was over something that scarcely existed. Even at the height of the so-called "rehabilitative era," a corrections department spending more than 2 percent of its budget on treatment was unusual. But the attack was taken up by liberals and conservatives alike -- many of whom felt that belief in rehabilitation, as Harvard's James Q. Wilson put it, "requires not merely optimistic but heroic assumptions about the nature of man." But as Berkeley criminologist Elliott Currie would later explain, "programs cited by Martinson and other critics as evidence that rehabilitation did not work were often not only underfunded and understaffed, but typically staffed by poorly trained and often unmotivated people. These early critics of rehabilitation made little effort to separate reasonably serious and intensive programs from those -- vastly more common -- that at best offered minimal counseling or tutoring to people who were otherwise allowed to languish in the enforced bleakness of institutions or in the shattered, dead-end communities from which they had come." The classic 30-year "Cambridge-Somerville Youth Study" is a premier example. In the Harvard-sponsored program begun in 1937, researchers followed 320 boys for 30 years. The boys were assigned to 10 "counselors" who had no training in mental health or psychotherapy and were told to do "whatever they thought best." Each youth was seen only five times annually during the early years of the project. Not surprisingly, the program had little effect on subsequent criminal behavior. Problems of Perspective Part of the problem in evaluating rehabilitation is deciding what constitutes success. For example, in studying the effectiveness of family therapy with hard-core delinquents (each having 20 or more previous convictions), one survey found that after 15 months, 60 percent of those in therapy had re-offended. However, 93 percent of the matched "non-therapy" control group re-offended. A medical procedure that suppressed symptoms in 40 percent of a group of chronically ill patients, 93 percent of whom deteriorate without treatment, would be seen as a virtual triumph. In corrections, however, such results are usually regarded as failure. Moreover, simply residing in some communities increases the likelihood of being labeled a recidivist. Nearly half (46 percent) of the boys in some areas will appear in juvenile court during their teen years. Among young black men in certain parts of the country, seven out of 10 can anticipate being arrested at least once. Though this may suggest failure, it may not measure individual criminal behavior. Indeed, among chronic delinquents the simple fact of rearrest may be less important than whether the young offender is winding down his criminal activity. But the biggest problem in getting a fair hearing for rehabilitation is that so many efforts have failed spectacularly. A team of researchers from the Academy for Contemporary Problems found that the "velocity of recidivism" among youthful offenders actually increased with each trip to a state reform school for rehabilitation. Rand Corp. researchers reported similar patterns among adults. Nonetheless, some theorists maintain that the very fact that a prison is dangerous and violent makes it rehabilitative. It's a variation on the "Scared Straight" theory. Unfortunately, repeated studies have shown that it doesn't work. The Incapacitation Gambit So, runs the presently popular notion, if we can't get a complete "cure," why not simply lock up all offenders? Simon Dinitz from the Academy for Contemporary Problems' "Dangerous Offenders Project" considered this Draconian option. He estimates that incarcerating every first-time felony offender for five years would likely yield no more than a modest 7.3 percent decrease in crime rates. But U.S. prisons (already overcrowded) would have to increase their populations 300 to 500 percent, entailing construction costs of $130 billion and increasing annual operating budgets from $12 billion to between $36 billion and $60 billion. And even that would not guarantee that crime rates would stay down for long. Those in prison are often replaced by others waiting in the wings (particularly among drug offenders). More ominously, such a policy would yield 3 to 5 million slightly more hardened ex-convicts dumped into the streets every five years. The most unusual case for incapacitation was made late last year by Richard B. Abell, an assistant attorney general in the Justice Department. Writing in Policy Review, and using figures compiled by a Justice economist, Abell concluded that we save $40 million annually in crime costs for every 100 offenders we incarcerate -- based on the extraordinary assumption that a typical offender commits 187 crimes per year at an average $2,300 per crime, or $430,100 annually. Calling these estimates "not merely wrong, but ludicrously wrong," University of California researchers Franklin Zimring and Gordon Hawkins noted that at a rate of 187 crimes per offender per year, putting a half-million more persons in prison would lower the number of crimes nationally by almost 50 million -- thus making the nation crime-free, since there are about 45 million crimes reported annually. By Abell's calculations, in fact, crime must have disappeared sometime in late 1985 as a result of the doubling of prison and jail populations from approximately 300,000 in 1978 to about 800,000 in 1986. Nonetheless, President Bush -- who pledged during the campaign to double the federal prison-building budget over four years -- has used the same argument. All this suggests that we are willing to invest large sums in variations on themes of retribution and deterrence. Yet Canadian psychologist Paul Gendreau and University of Ottawa sociologist Robert Ross, citing sophisticated new mathematical analyses of the data on rehabilitation, concluded that "the (substantiated) claims for effective rehabilitation of offenders far outdistanced those of the major competing ideology: applied deterrence or punishment." Actually, Something Works As early as 1976, a Rand Corp. report had suggested that the "nothing works" conclusion was probably premature. Three years later, a National Academy of Sciences panel concluded that "when it is asserted that 'nothing works,' the panel is uncertain as to just what has been been given a fair trial." And now, in their latest survey of the rehabilitative literature, from 1980 to 1987, Gendreau and Ross found "reductions in recidivism, sometimes as substantial as 80 percent had been achieved in a considerable number of well-controlled studies. Effective programs were conducted in a variety of community and (to a lesser degree) institutional settings, involving pre-delinquents, hard-core adolescent offenders and recidivistic adult offenders, including criminal heroin addicts." The literature of the '80s demonstrates that a number of techniques can reduce recidivism among both property and violent offenders. These include substance-abuse treatment (combining intensive counseling with drug screening), family therapy, individual therapy stressing support rather than pathology and punishment, and -- particularly with young offenders -- assigning "advocates" to work with individuals on a daily basis, including crisis intervention at odd hours. In Massachusetts, Harvard researchers found that reconvictions fell among older former reform-school youth when a range of such alternatives was available. In those regions of the state where no such array existed, recidivism remained the same or increased. Educational programs for hard-core adult offenders have also shown promising results. Inmates of a Canadian federal prison, many with long and serious criminal histories, were assigned randomly either to normal prison routine, or to a special humanities program stressing individual tutoring using Socratic dialogue. In a report prepared for the Canadian government, psychologist D.J. Ayers and his colleagues found that after 20 months of post-prison follow-up, the recidivism rate of those in the program was 14 percent as compared to a 52-percent rate for those randomly assigned to prison routines. Discovering what works is less a matter of deciding on a specific treatment technique than of creating programs that are intensive, taken seriously, last a reasonable period of time and focus on high-risk offenders. (In fact, programs directed at low-risk offenders can sometimes be counter-productive if they are allowed to pick up antisocial skillls and attitudes from higher-risk persons.) Canadian psychologists D.A. Andrews and J. Keissling found that effective therapy promoted prosocial attitudes, rewarded non-criminal pursuits, made use of a wide range of community resources, taught skills for handling relapse and treated the offender with respect and empathy -- many of the very qualities that characterize effective psychotherapy with non-offenders. Ironically, even Martinson himself changed his mind on the efficacy of rehabilitation. In a 1979 article in the Hofstra Law Review, he wrote that "startling results are found again and again . . . for treatment programs as diverse as individual psychotherapy, group counseling, intensive supervision and what we have called individual help." The man who started it all had come full circle. But by then no one was listening. And apparently they still aren't. On Jan. 18, the U.S. Supreme Court confirmed the abandonment of rehabilitation. In Mistretta v. U.S., the Court upheld federal sentencing guidelines which all but remove rehabilitation from serious consideration. The dissonant reverberation from a decade earlier has become the national anthem. As a result, federal prison populations are expected to double. Thus contemporary corrections theory offers a choice between equally unattractive extremes: ineffective probation/parole or debilitating prisons. Finding help is akin to asking a doctor for headache relief and being told there are only two treatments -- an aspirin or a lobotomy. Harsher sentences, warehouse prisons and an ideology which militantly rejects the idea of salvaging offenders are the rule of the land. Meanwhile, violent crime surges. We must now wait for the swing of the pendulum. It may be a long wait. Hard-Core Success IN DETROIT and Boston, successful programs for violent young offenders begin with a brief period of structured therapy in a locked residential setting, followed by intense and long-term community supervision. New York researchers Jeffrey Fagan and Eliot Harstone discovered that these programs cut recidivism significantly because they fit the therapy to the individual needs of each youth, provided extensive social networking and "continuity of care" and were perceived by the youths as offering real opportunities for success. Similarly effective results have come from Massachusetts where, since the closing of state reform schools in 1973, a private non-profit program called "Key" has offered intensive "outreach and tracking" for high-risk delinquents. Key trains mostly college-age paid staff to work individually with the youths 10 to 50 hours per week in their own homes and communities, usually in the evening or on weekends -- the times when most youngsters get in trouble. Key workers are not neutral, but are advocates for their charges. (When monitoring is done by more disinterested juvenile-court personnel, the results have been far less successful.) Key advocates negotiate the school system, cajole human-service bureaucrats, intervene when a youth strays, deal with police and encourage legitimate activities. While in the program, 92 percent of these high-risk youth remain offense-free. And a program for delinquents at four sites in Michigan -- in which college students trained as paraprofessionals gave six to eight hours of counseling per week to each youth -- reduced recidivism rates by nearly a third. All this suggests that the most successful agents in rehabilitation may not be those in the roles familiar to American psychiatry, psychology and social work. Such professionals typically have preferred to work with the "most likely to succeed" -- the most articulate and socially skilled offenders who are least in need of professional care. Rehabilitation directed at the more problematic cases requires a new kind of worker who is willing to reach out, pursue and even advocate for offenders who are unlikely to fit the 9-to-4 regimen of the traditional mental-health professional. A fitting epilogue to the rehabilitation debate can be found in the research conducted in 1987-88 on New York's successful "Stay'n Out" therapeutic community drug abuse treatment program -- another model with an extensive aftercare component. Both male and female drug addicts showed dramatically lower arrest rates than control groups. The research monograph was coauthored by Douglas Lipton, senior author of the 1975 survey which Martinson claimed showed that "nothing works." Lipton is now a leading advocate of rehabilitation in corrections. Jerome Miller is director of the National Center on Institutions and Alternatives.