American consumers and policy makers suffer from a lack of reliable information about college tuition and financial aid for students, two economists said last weekend at the annual meeting of the Allied Social Science Associations.

"We all hear a lot of stylized facts about tuition," said Caroline Minter Hoxby, a professor of economics at Harvard University. "But these things that we think we know are not actually all that accurate."

At a panel discussion on Sunday, Ms. Hoxby and Catharine B. Hill, of Williams College, reported on new research programs that are designed to illuminate the real-world dynamics of college costs.

One project, led by Ms. Hoxby, is an attempt to reliably measure changes in tuition over time. The most commonly cited tuition statistics, like the College Board's annual "Trends in College Pricing" report, have gross methodological flaws and seriously overstate the annual rate of tuition growth, Ms. Hoxby argued. She estimated that the College Board's report overstates the growth of tuition "by at least 30 percent -- by 29 percent for four-year private institutions, by 15 percent for four-year publics, and so dramatically for two-year publics that it's not really worth giving you the numbers. It's more than twentyfold."

One of the authors of the College Board report strongly denied that the report seriously overstates tuition growth. In an telephone interview on Monday, Sandy Baum, a professor of economics at Skidmore College and a senior policy analyst at the College Board, conceded that the report is imperfect, because many institutions do not reply to the College Board's survey. But she said that there is no reason to believe that the gaps in the data systemically bias the tuition figures upward.

In her presentation, Ms. Hoxby said a more reliable measure of tuition growth would weight colleges by their size. "Ohio State has fifty-some-thousand students," she said. "Colby College has some tiny number of students. They're both weighted equally" in the College Board's reports and other commonly cited tuition studies, she said. "This just does not make sense."

By analogy, she explained, it would be foolish to assess car prices by simply averaging the $18,000 cost of a General Motors car with the $200,000 cost of a Rolls-Royce. An accurate portrait of the automobile market must take account of the fact that General Motors sells many more cars than Rolls-Royce does. Economists make such weighted adjustments when they assess car prices, but analysts of college tuition often do not.

She also argued that statistics should rely not on a college's official tuition, but on the prices that students actually pay, after need-based financial aid is taken into account. "We would think it was pretty absurd if the government's automobile-price series was based on the sticker price and not on actual prices paid," she said.

Once you account for institution size, Ms. Hoxby said, the growth in both average tuition and median tuition since 1968 appears much more moderate. The growth in tuition has been concentrated at smaller colleges -- and, in particular, at the most expensive institutions. If you look at the price distribution of colleges, she said, "all the action is at the 80th percentile and above. At the 70th percentile and below, there has been only very modest tuition growth."

Referring to recent Congressional proposals to regulate tuition, the professor argued that education policy should not be driven by the widely publicized prices of a few elite colleges.

"This is where I think people lose perspective," she said. "These colleges may be very important to us, they may be very important to the people in this room, they may be very salient to congressmen and reporters. But realistically, this is a tiny share of the American college market. These prices are not relevant to the experience of the typical student."

Ms. Hoxby's project began as a digression from her main line of research. She is writing a book on higher education and wanted to cite data about changes in tuition over the past three decades. She initially believed that doing so would be a simple matter of quoting official government documents. But, she said, she discovered that neither the federal government nor any private entity maintains tuition data that are as statistically sound as, for example, the U.S. Bureau of Labor Statistics' annual price reports for such goods as cars, housing, and food.

"I got more and more annoyed about the fact that there wasn't a good series," she said. "The policy debate depends on people understanding the truth about college tuition. Your typical state or Congressional legislator just does not know this."

Ms. Hoxby said it would be desirable for the federal government to maintain reliable data that would help particular groups of students -- those who are high-achieving or financially needy, or who are from a particular region -- make intelligent choices on where to attend college.

"Although we are consumers of colleges, college is really an investment in human capital," she said. "So you shouldn't deal with it as if it were a simple price index, and we were just going to consume the thing called 'college.' It's really about investment, so you should always compare college costs to returns." Such returns might be measured, for example, by the average lifetime earning of a particular college's alumni, or by how quickly a college's alumni pay off their student debt.

The other project presented on Sunday is an attempt to identify both the "sticker price," or official tuition, and the actual, or net, tuition paid by undergraduates at 28 selective private colleges and universities. The study includes all eight Ivy League colleges, along with other prominent institutions, like Amherst College, Duke University, Stanford University, and Williams College.

Ms. Hill, a professor of economics at Williams, began with the familiar observation that many students at such institutions receive at least some need-based aid and therefore pay considerably less than the official tuition. Forty-five percent of the 108,721 students in Ms. Hill's study received need-based aid in 2001-2. They paid, on average, 47 percent of the sticker price. Financially needy students -- those in the lowest quintile of national household income -- paid, on average, just 22 percent of the sticker price.

"The press often reports the sticker price of schools like ours relative to U.S. median family income," said Ms. Hill, "and suggests that that's an indicator of what it costs to attend one of these schools. But, in fact, a kid who actually comes from a family with the median income isn't asked to pay full price. ... If you look at our sample, what we ask a kid at the median family income to pay is $11,557. Relative to median family income, that is 23 percent. The number that frequently gets reported in the press is the full sticker price relative to U.S. median family income, which is around 66 percent."

As a result, Ms. Hill said, many high-achieving high-school students believe incorrectly that elite colleges are financially out of reach.

Things look less equitable, however, if one looks at net tuition as a fraction of family income at specific levels of neediness or affluence. In 2001-2, the 28 colleges in Ms. Hill's study asked students from families in the lowest income quintile to pay, on average, $7,552, which is 49 percent of the median annual family income ($15,400) in the lowest quintile. Students in the top income quintile who received need-based aid paid, on average, $23,690, which is just 21 percent of the median annual income in the highest quintile.

Another member of the panel, Eleanor P. Brown, a professor of economics at Pomona College, said the net-tuition measure might not be as regressive as it appears. She said she suspects that needy students typically pay considerably less than $7,552, after non-need-based forms of support, like National Merit Scholarships and athletics scholarships, are added to their aid packages. Merit-based aid was not included in this phase of Ms. Hill's study.

Ms. Hill emphasized the substantial variety among the 28 colleges in her study. Four have progressive financial-aid structures, in which students from higher-income families are asked to pay higher fractions of their family income. Seven of the colleges have regressive patterns, in which students from lower-income quintiles are asked to pay higher fractions of their family income. The other 17 colleges lie in between, asking students from each quintile to pay a roughly proportional fraction of family income.

What determines which of the structures a college adopts? "Wealth seems to be an important part of the answer," Ms. Hill said. Colleges with larger endowments are statistically more likely to have progressive pricing structures.

"If you have more money, you can pay more aid," she said. "But, in fact, what we think is going on is a more indirect effect. The wealthier schools are the more selective schools," and the students they choose tend to come from higher-income families. "So you end up with a student body that has very few needy kids in it, and you can afford to be quite generous with kids in the first and second quintiles."

Ms. Hill's work was conducted under the auspices of the Williams Project on the Economics of Higher Education. Her paper was written with Gordon C. Winston, a professor of political economy at Williams, and Stephanie Boyd, a research associate there.