Peter Anderson, a freshman at Johns Hopkins University in Baltimore, may have made a costly mistake: He told the school that he planned to be a premed major.


The problem is, Johns Hopkins already has plenty of premed majors. It wanted more humanities majors. And so in an experiment last spring, it quietly offered fatter financial-aid grants to incoming humanities majors than to most of their premed counterparts. While Peter is getting $14,000 a year, he might have snared about $3,000 more if he planned to major in, say, art history.


"What you're telling the kid is, lie to get into college," says Peter's astonished mother, Joan Anderson, a New Kingstown, Pa., homemaker, after she is told of the practice by a reporter.

Johns Hopkins's bold experiment is being tried out at colleges all over the country. At these schools, grants no longer are based overwhelmingly on a student's demonstrated financial need, but also on his or her "price sensitivity" to college costs, calculated from dozens of factors that all add up to one thing: how anxious the student is to attend. The more eager the student -- and Hopkins is high on the list for premed majors like Peter -- the less aid they can expect to get. Although students and families awaiting word of college admissions this week aren't being told, these colleges are employing some of the same "yield management" techniques used to price and fill airline seats and hotel rooms.


The statistical models, which have become widespread only in the past few years, go by innocuous names like "financial aid leveraging." But they are quietly transforming the size and shape of student bodies in all sorts of ways, some of which are alarming educators.

A sampling:

-- The Johns Hopkins model -- which the school isn't currently using but may try again in the future -- suggested slashing aid to some prospects who came for oncampus interviews. The reason: Those students are statistically more likely to enroll, so need less aid to entice them. A school official now denies putting that part of the model into practice.


-- At Pittsburgh's Carnegie Mellon University and other schools, eager freshmen accepted through the early-admissions program can end up with less financial aid than comparable students who apply later. "If finances are a concern, you shouldn't be applying any place early decision," says William F. Elliott, vice president for enrollment management.


-- For the current school year, St. Bonaventure University in Bonaventure, N.Y., gave the poorest of its top-ranked prospects just over half the grants they needed -- but gave more-affluent prospects more than three times their need. The result: A more-affluent student body, since 75% of the wealthier students decided to enroll, while only one in 11 of the neediest did so.


"This is an ugly, ugly business," says Walter C. Cathie, dean for enrollment management at Wabash College in Crawfordsville, Ind., who says he left a financial-aid post at Carnegie Mellon last year partly because of ethical concerns. Even two top Hopkins aid officials admitted at an industry gathering that the public-relations consequences if word leaked out could be "scarifying."


The schools argue that the new financial-aid engineering is necessary at a time when tuitions are high and many private schools don't have enough aid to go around. The complex computer modeling, they say, is designed to help them get the kind of students they want without giving away too much of their own money -- since aid grants are, after all, a form of price discounting.


About 60% of the nation's 1,500 private, four-year colleges now use statistical analysis "in some form" to dole out aid, estimates Steven T. Syverson, dean of admissions and financial aid at Lawrence University in Appleton, Wis. "Maximizing revenue is the buzzword of educational administrators these days," explains Jon Boeckenstedt, St. Bonaventure's dean of enrollment management. He concedes "there are some inequities certainly operating" at his school, but adds, "I would challenge anyone to come up with a system that's always fair and always equitable."


Ivy League colleges and others with big endowments have enough financial-aid funds to meet all their students' needs, and so generally don't rely on such Darwinian methods. Critics say other colleges shouldn't do so either. Financial-aid leveraging, they say, is discriminatory, and acts as a second tier of admissions. At its most insidious, critics say, it can eradicate "need-blind" admissions altogether. After all, what does it matter if the admissions process is need-blind, if a college doesn't give the neediest enough aid to enroll?


"I think it's terrible," says Hugh Chandler, chairman of the guidance department at Weston High School in Weston, Mass. He says colleges should base their aid decisions on "what makes sense for a kid, rather than how are we going to get the biggest bang for a buck." Enrollment management of this type violates the "good practice" guidelines of both the National Association of Student Financial Aid Administrators, or Nasfaa, and the National Association of College Admission Counselors, both groups say. But neither has enforcement power -- and Nasfaa says it is considering changing its code so the practice won't be considered unethical anymore.

Colleges, of course, have always offered their most attractive aid packages to their hottest prospects, such as athletes or especially gifted students. But the sophisticated computer analyses that so many schools use now didn't start until about 15 years ago, and didn't become widespread until 1992.


That year, Congress liberalized the federal formula for calculating financial need, making most families eligible for more aid. Unfortunately, the government didn't provide extra federal funds to pay for all that extra aid. Colleges were forced either to boost their own financial-aid budgets, or to offer aid more selectively. Many chose the latter path, out of necessity.


To achieve their ends, these schools still start by calculating a student's demonstrated "need" -- determined by family income, assets and debt. Instead of stopping there, though, some schools then factor in dozens of variables that affect a student's propensity to attend the college once he or she is accepted. The higher the propensity, the less financial aid the student may expect. Factors can include a student's home state, ethnic background and area of study, and who initiated the first contact with the school.


"Those who have the most interest in the school are going to be less price sensitive," explains Stephen H. Brooks, a Waltham, Mass., consultant who has performed statistical analyses for about two dozen schools, including Johns Hopkins, New York University, Rochester Institute of Technology, and Hobart and William Smith colleges.


The mathematical models are customized for each school. Drexel University in Philadelphia, for example, used a model that accurately predicted more business majors would enroll if it sweetened their offers, according to Donald G. Dickason, who recently retired as head of enrollment management. Certain other students were offered less-generous aid packages to make up the difference, he acknowledges, but he won't say which ones. His successor, Gary Hamme, didn't return calls to discuss whether the model is used now.


Creating the models is a delicate business, and a host of consultants have sprouted up to tackle the job. A brochure from the National Center for Enrollment Management in Littleton, Colo., promises an analysis that "links ability to pay with willingness to pay" and "packaging strategies that directly support your goals for both new and returning students." Thomas E. Williams, president, says his clients last year saw an average net tuition-revenue gain of $474,000 as a result of financial-aid leveraging.


A pamphlet from Dr. Brooks, the Waltham, Mass., consultant, questions whether colleges are "too generous, too stingy or just right" with their institutional grants. "Did you overspend to get students who would have matriculated with lesser aid? Did you underspend and lose students who would have come with more support?"


Dr. Brooks, whose models sometimes suggest less generosity to early-decision applicants, offers this advice to families: "If you go into the showroom and say you want to have that red Corvette, they're not going to cut the price much. I would say being a little cagey would be helpful."

Johns Hopkins turned to Dr. Brooks last year when it wanted to increase its humanities enrollment by 20% and reduce its overcrowded premed program. At an Nasfaa conference in San Antonio last summer, a Hopkins official explained how the school achieved its goal by using an econometric model, a mathematical analysis that tries to predict down to the dollar what it will cost to persuade a student to enroll.


The model suggested that Hopkins should offer $3,000 more to prospective humanities students with Scholastic Assessment Test scores over 1,200 and relatively low financial-aid needs; the extra money would increase enrollment probability by nine percentage points. One group of students, the model suggested, shouldn't get the extra cash: those who had had campus interviews, a step Hopkins itself "strongly" recommends. Campus interviewees already were about 9% more likely to enroll at Hopkins than other prospects, so the cash incentive wasn't necessary, concluded the analysis, which was based on two years of past enrollment data.

The model also suggested cutting by $1,000 the aid offers to most premed campus interviewees with SAT scores below 1300. Those students were already hooked, so "it wouldn't knock that many students out. But it would increase the net revenue from this group," explained Robert J. Massa, the school's dean of enrollment management.


"So let's look at the results," said Dr. Massa, as he displayed a bar chart on an overhead projector. "In fact, what we wanted to happen did happen." Humanities students increased "by exactly 20%" while the premed group dropped "by about 10" students, he said. He stressed, however, that the analysis was "just one tool," not "the sole path."


During the presentation, which was taped, Dr. Massa acknowledged that the model raised ethical questions. When an audience member asked, "How upfront are you with families about what you're doing?" Dr. Massa conceded the school hadn't told them about its formulas. He said colleges should come clean "as we become more and more sure of what we're doing . . . ." Yet he added, "They [parents and students] don't need to know the specifics of, `If you come for an interview you're not going to get financial aid,' for example."


Interviewed later, Dr. Massa emphasized that the model has only been used once so far, and that Hopkins didn't follow it exactly. He denied that Hopkins penalized students who had campus interviews. The school actually offered about $3,000 extra in grants to all prospective humanities majors except early-admission applicants, he said. And rather than cutting natural-science majors' aid outright, it gave most of them the college's least-attractive aid package, with about $1,500 less in grants and $1,500 more in loans than the best package.


Carnegie Mellon, considered a pioneer in statistical modeling, has taken the process further. Its enrollment vice president, Dr. Elliott, wrote a doctoral thesis 22 years ago on maximizing net tuition revenue. Eight administrators meet for several hours each week to tinker with formulas and review the latest spreadsheets that attempt to predict "yield" -- the percentage of accepted students in different groups that will actually enroll, if money is taken from one group and given to another. Without such work, Dr. Elliott says, "I'd have an institution full of engineers and computer scientists and I wouldn't have anybody in arts and design."


An internal "Awarding Strategy" for the university's popular computer-sciences school for the 1995-96 academic year suggested offering some top-ranked students almost five times their demonstrated need in cash grants if they qualified for $4,000 or less in aid. For similarly ranked students who required more than $20,000 in aid, the model suggested offering 75% in grants. Meanwhile, no grants were budgeted for the 142 students ranked at the bottom -- fully 60% of accepted students -- regardless of need.


Dr. Elliott declined to say how the award strategy was carried out, but Mr. Cathie, the former Carnegie Mellon financial-aid administrator, says he believes it followed the model's recommendations.


Carnegie Mellon also takes an aggressive approach toward its competition. After admitted students receive their financial-aid offers in the spring, they are invited to fax the school any better offers they receive from other colleges. Carnegie Mellon sets aside more than $250,000 in its budget for them in what is known internally as the "Reaction Program." For desirable students, the school generally meets competing offers.


Of course, early-admission candidates at Carnegie Mellon don't benefit from the program, since they agree to withdraw all other applications upon acceptance. Nor does the university disclose this fact to them in the application process. Early-admission applicants also sometimes receive less generous financial-aid offers because their packages are based on the previous year's statistical models; regular applicants might receive "hundreds" more, depending on what the new model suggests, Dr. Elliott says.


Carnegie Meltem sometimes troubles Dr. Elliott "These are gut-wrenching decisions," he says -- but he feels there is little choice. "Obviously, I don't have enough money to be as generous as I might like to be," he says. "I could make it very fair -- and be out of business."