(FORTUNE Magazine) – CHANGING human behavior is like reshaping concrete. The material is stubborn. Over the past couple of decades government agencies as well as corporations and private foundations have spent billions trying to blunt destructive habits that create grim casualty lists: four million cocaine and heroin addicts, one million pregnant teenagers a year, and 700,000 high school dropouts. Can anything hold down the mounting toll? The answers come at a maddeningly slow pace. A hard-pressed drug clinic would rather spend money to keep a program rolling than to study whether its graduates remain clean. Promising ideas don't circulate fast enough, and failed ones persist, partly because theorists are inclined to stick with treatments that ought to work -- but often don't. Though evidence from the field is incomplete, enough exists to sketch some general conclusions. There are four basic approaches to changing behavior: education, punishment, financial incentives, and -- often the last resort -- treatment. Under the right circumstances, all work some of the time and none work all of the time, despite the attempts of advocates to portray one or another as a panacea. In all four cases programs for the very young will likely produce the best results. New York City stumbled when it spent $120 million in the late 1980s to persuade 150,000 wavering students to stay in high school. The city hired extra guidance counselors, offered wilderness training to some youngsters to build their confidence, and rigged up an automatic telephone dialing system that rang the truants' homes and reminded them to come to class. The effort failed; the dropout rate remained at 28%. The program almost certainly came too late. The 16-year-old who drops out in New York probably started to make that decision, unconsciously, a decade before. Dr. Sheppard G. Kellam, chairman of the mental hygiene department at Johns Hopkins School of Public Health, has tracked hundreds of youngsters from 6 to 16. His statistics show that those likely to drop out -- and become heavy drinkers and drug users as well -- can be spotted early. Most are what he calls the shy-aggressive type, belligerent loners who emerge from their solitude mostly to strike a schoolmate or defy a teacher.

Working with the Baltimore public school system, Kellam is trying to figure out how to reform these tiny classroom bullies before they cause grown-up trouble. Some are put in classes where teachers divide students into teams. Privileges are awarded on the basis of how well the team, rather than any individual in it, behaves. The idea is to reform delinquents through peer pressure. Preliminary results are mildly encouraging. Some 20% of the children described by teammates as ''mean'' at the beginning of the year were rated ''nice'' at the end. But no classroom remedy is likely to work unless the parents or grandmother or big sister backs it up at home. As Harvard psychologist Jerome Kagan puts it, ''It's hard to treat a child for diarrhea if you send him back to a swamp every night.'' So Kellam's next daunting step will be to get the guardians to cooperate. They will likely have to alter their own behavior, even more deeply embedded than that of the children. EDUCATION is a potent force, though proponents are inclined to exaggerate its power. Senator Edward Kennedy (D-Massachusetts), for one, insists that instruction can inoculate children against the drug epidemic. Not likely. Many schools don't even understand what it is they are supposed to teach. Simply laying out the physiology of drugs or alcohol or sex doesn't help much; it is familiar to youngsters anyhow. It is better, say experts, to focus on what you might call the social aspects surrounding such activity. What the 15-year-old girl doesn't know is how to bring up the subject of contraception with her new boyfriend. ''For many kids it's easier to jump into bed than discuss sex,'' says Joan Lipsitz, a director at the Lilly Endowment, a private foundation in Indianapolis. ''Sex isn't a difficult act for them; talking is.'' The boy of 13 doesn't know how to turn down marijuana when it's offered by the most popular 16-year-old in school. One drug program, Project Star in Kansas City, has produced measurable results in helping youngsters cope with such situations. Begun in 1984 with the support of Ewing Kauffman, founder of Marion Laboratories and owner of the Kansas City Royals, it relies on role playing and other training techniques familiar to business. Kauffman insisted on market research of sorts. Says he: ''I'm a penny pincher, so I wanted to test this program just as I would a new pharmaceutical.'' Some 15,000 sixth- and seventh-graders were divided into two groups: The experimental group got an hour a week on substance abuse. Rather than giving lectures, teachers encouraged students to discuss their curiosity and fears and develop their own standards, which sometimes turn out to be remarkably conservative. Thus, no matter what the pressures outside the classroom, the students would know that at least one community shared their values.

The other group of students, intended as a control, had no special instruction on the subject. University of Southern California researchers periodically surveyed and tested both groups. The experimental group used all substances less frequently. Half of the control group drank alcohol, compared with 35% of the experimental; 32% smoked cigarettes, vs. 24%; 20% used marijuana, vs. 14%. Getting accurate information about what their peers were doing proved especially important. At the start the youngsters in the experimental group assumed that two-thirds of their classmates had tried forbidden fruit. Fewer than a third actually had, which greatly reduced social pressure on the abstemious. Each student was encouraged to develop a particular strategy to resist peer pressure, then practice it repeatedly in minidramas with classmates, much the way a business executive might rehearse an upcoming confrontation with a difficult client. How to resist the big man on campus when he offers marijuana? Maybe the BMOC has no clear idea of whether he is testing his junior or simply making a perfunctory gesture, so the 13-year-old might demur gracefully with a terse ''No, thanks.'' (Take a small bow, Nancy Reagan, but the ''Just'' in the ''Just say no'' campaign makes it sound too easy.) Indianapolis has copied the plan, and Washington, D.C., is testing it in a few schools. Some of the elements that helped it work in Kansas City, however, cannot be easily duplicated. Kauffman's prominence as owner of a major league baseball team helps too, by attracting media attention. The Midwestern city has its social and economic troubles (9% of the population lives below the poverty level) but not on the scale of Chicago or Los Angeles. What happens to peer pressure, for example, when the 13-year-old discovers that a critical mass of his classmates are indeed substance users? One possible solution: Start the program at a lower grade. PUNISHMENT can be a powerful deterrent, psychologists say, but only when the wayward perceive it as swift and sure. That greatly limits its effectiveness in a society that cherishes due process. An impressive exception is the campaign against drunken driving, where punishment has been far more effective than either treatment or education. Attacking the problem as a treatable disease misses the mark. It is not the archetypal alcoholic who poses the greatest danger on the road. Says James L. Nichols, an authority on drunken driving at the Department of Transportation: ''A 20-year-old male who has had eight beers is more likely to drive recklessly than a 40-year-old with a severe alcohol problem.'' Mandatory safe- driving classes don't help much either. What does work are well-publicized campaigns in Florida, California, and 18 other states to thwart drunken drivers by suspending their licenses for 30 days or more. At random police checkpoints, drivers have to pass Breathalyzer exams or perhaps simple arithmetic quizzes. Those who flunk must hand over their licenses, although police give temporary permits to allow a chance for appeal. That's about as sure and swift a punishment as the Constitution allows, but its effectiveness is not as straightforward as lawmen would like to think. Says Nichols: ''Most of those who lose their licenses continue to drive. But because they're worried about getting caught, they drive carefully.'' Compared with drunken drivers who are merely warned or put through education programs, drivers whose licenses are suspended have 40% to 50% fewer crashes for as long as five years after their arrests. Some clinics report they can help smokers quit with shock therapy. In a typical setup the smoker, who sits in a room with a wire attached to one wrist, gets a small jolt after reaching for a cigarette. Away from the clinic the person snaps a rubber band on the wrist after lighting up to recall the punishing shock. When the consequences don't have to be faced immediately, punishment is far less fearsome. Teenagers are careless about using condoms, despite the risk of AIDS. Ruth Wooden, president of the Advertising Council, says interviews with teenagers show that what worries them about tobacco is not so much emphysema at 60 as bad breath on Saturday night. FINANCIAL INCENTIVES, even small ones, are surprisingly effective. But, as with punishment, the payoff must be quick. In 1984, Planned Parenthood of the Rocky Mountains in Denver offered girls under 16 who had previously had babies, abortions, or miscarriages $1 a day not to get pregnant again. Once a week they come in to pick up the wages of virtue. Though they don't have to, many hang around to talk about baby care or their own health or to gossip. Says Dr. Margaret LaTourrette, a Planned Parenthood board member: ''They wouldn't come without the money. Some buy diapers, some buy hair spray.'' On average half of teenage girls who get pregnant once do so again as adolescents. Over the past year only two of the 57 girls in the Rocky Mountain program have become pregnant. That is a very small sample on which to base big conclusions, but the program is so inexpensive that it seems worth trying elsewhere. Dr. Larry Culpepper, who runs the Blackstone Valley Perinatal Network in Rhode Island, attracts poor pregnant women to his program by offering them a chance to win a lottery of from $25 to $500. Traffic is up 30% since the payoffs began more than a year ago. ''These women want healthy babies just like other mothers,'' says Culpepper. ''They don't know help is available, and they don't read pamphlets. But if someone wins $100, the word gets around the housing project fast.'' Even if only one woman carries her baby to full term as a result, the lottery will justify its grants, $15,000 from the March of Dimes and $5,000 from the federal government. A premature baby costs an average of $22,000 more to care for than one carried to full term. And taxpayers usually pay the bill. Financial penalties work in at least one case better than many economists have assumed. The conventional wisdom about smoking is that the habit is not sensitive to changes in prices. But according to a study by the National Bureau of Economic Research, smoking decreased between 1955 and 1985 in states where excise taxes rose. While New York City attempted to coax kids to stay in school, Wisconsin threatened their mothers. With Learnfare, a program developed under Republican Governor Tommy G. Thompson, recipients of aid to families with dependent children (AFDC) forfeit a share of their benefits if their teenage children skip school. The program appears to be bringing children back to the classroom, but it isn't clear whether the returnees are studying or just filling seats. Economics by itself doesn't seem to hold a drug addict's attention for long. Dr. Roger Meyer, head of the department of psychiatry at the University of Connecticut, recalls a treatment program for heroin addicts he helped direct in the mid-1970s. As long as the patients were in the hospital and under supervision, they performed well for rewards, such as $4 for writing a job resume. After they were released they were supposed to show up daily at a local pharmacy, swallow a potion that blocks heroin, and collect $1. Many fell away. Those who stuck with the program tended to be the ones with strong family ties. TREATMENT follows when all else fails, but how effective is it? Most drug clinics don't keep track of their discharged patients closely enough to answer the question statistically. The anecdotal evidence suggests that the failure rate is high. Addicts generally relapse at least once or twice, says Dr. Herbert D. Kleber, deputy director of demand reduction for the Office of National Drug Control Policy. The trouble is, an addict doesn't enter treatment the first time to get rid of his habit. Rather, says Kleber, ''he wants to get back to that honeymoon when the drug felt great and he could control it.'' Nonetheless, like smokers, a lot of drug addicts and alcoholics keep trying to reform. Dr. Roger Weiss, who runs the alcohol and drug treatment program at McLean Hospital near Boston, says that, contrary to the cliche, the most promising candidate is not one who has hit bottom but one who is just an arm's length away. Says Weiss: ''He is the guy who thinks about selling his mother's wedding ring but hasn't done it yet.'' The combination of a downside risk in continuing and an upside opportunity in stopping seems to provide the best incentive. Government and private agencies could do a much better job of directing drug abusers to the appropriate place. Says Kleber: ''Where an addict gets treatment depends very much on which door he happens to knock on. It is senseless to put someone with no skills into a 28-day rehabilitation program. He has to be habilitated.'' Learning a trade, as well as the discipline to show up for work and take orders, could require as much as 18 months. Also, says Kleber, states should require hospitals and other treatment centers to achieve a minimum success rate in order to get government funding; perhaps one-half to two-thirds of their patients should be drug free a couple of years after release. Programs that take the toughest cases could have more lenient standards. Partisans of all strategies would do well to acknowledge that none are more than moderately effective. Programs that don't meet standards should be junked, no matter how compelling the educational or punitive theory behind them. And programs that do work should be analyzed more critically to find out what it is that makes them effective. Even a modest improvement in suicidal human behavior would be reason for the rest of society to cheer.