The wide gaps in educational attainment between people from low-income backgrounds and wealthier ones tend to lead policy makers to a logical conclusion: It's about the money. If low-income students are enrolling in college at significant lower rates than others, they argue, it's essential to increase spending on need-based financial aid and make sure the students know the money is out there. If needy students drop out of college at higher rates, then surely money -- or lack of it -- is the key factor there, too.

But a new study questions that thinking. The study, conducted by researchers at Berea College and the University of Western Ontario and available (for $5) through the National Bureau of Economic Research, examines the extent to which "credit contraints" -- a lack of access to loan funds -- factored into the decisions to drop out of college by a group of low-income students who had had most of their direct college costs paid for.

The answer: Not very much. The researchers conclude that more than 80 percent of the students who dropped out had done so for reasons other than financial ones. "The large majority of attrition would remain under the generous policy in which direct [college] costs are zero and students are given access to loans" that could be used to pay their other costs of living.

That doesn't mean that money doesn't matter, says Todd Stinebrickner, an associate professor of economics at Western Ontario, who co-wrote the study with his father, Ralph, a professor mathematics and computer science at Berea, whose students were the subjects. But the implication of the study is that "there's other stuff that's important, too," he says. "While it's very tempting to view a relationship by family income as being due to money per se, if you do, you might ignore a lot of other things. And for policy makers, if you think it's one thing and it's really something else, you might get the wrong policy."

Stinebrickner is careful to say that the study does not aim to undermine in any way the significant efforts that are being undertaken to make college more affordable for low-income students, such as the push, after several years of flat federal funding, to raise the maximum Pell Grant. But what's striking about the students in the NBER study, he and others note, is that they're at a college, Berea, that, because of its distinctive religious and service mission, charges no tuition and largely covers students' other direct educational costs.

The fact that even there, the proportion of first-year students who drop out by the start of the second year is about 23 percent, suggests "it's got to be something else" other than financial need or lack of access to borrowing.

Exactly what those other things are, Stinebrickner acknowleges, the research so far has not uncovered (future studies on Berea's rich longitudinal database are planned). But possibilities include such things as academic struggles, their satisfaction in college, how comfortable they feel in the environment, or just a feeling of "mismatch," he says. Each of those factors may require different tactics or solutions, and if college officials or policy makers are focused entirely on financial issues, "they may miss other things," Stinebrickner says, that would require examining campus culture, for instance.

The important finding out of the study for policy makers, says Christopher Avery, the Roy E. Larsen Professor of Public Policy at Harvard University, is that when trying to figure out "what can we do to help students with difficult family situations in terms of income, the natural tendency is to say, 'Let's throw some more financial aid at them.' But that may only be addressing some portion of the problem."