(AP) - Four years ago, two of the most influential researchers in higher education dove into a huge pool of data hoping to answer a bedeviling question: Why do so many students who start college fail to graduate?

They report their findings in a book out Wednesday, and perhaps the biggest is this: Students aren't aiming high enough, settling for less selective schools they imagine will be easier, but where in fact they're more likely to drop out before earning a degree.

In "Crossing the Finish Line," William Bowen and Michael McPherson, former presidents of Princeton University and Macalester College, along with researcher Matthew Chingos, chime in on what many experts consider American higher education's greatest weakness: college completion rates. By some measures, fewer than six in 10 entering college students complete a bachelor's degree, among the worst rates in the developed world.

The latest findings may surprise those caught up in the well-publized admissions frenzy at high-end colleges who assume all students push for the most selective school they can find. But the authors focus on the phenomenon called "undermatching" - the surprisingly large number of well-qualified high school seniors with credentials to attend strong four-year colleges, but who chose other options instead - less selective schools, two-year colleges, or no college at all.

They may have had their reasons, such as staying close to home or lack of money (though more selective schools aren't always pricier). But the authors argue bigger factors are "inertia, lack of information, lack of forward planning for college, and lack of encouragement." The data suggest low-income and minority students, and especially those whose parents don't complete college, are especially susceptible.

For instance, examining 1999 North Carolina high school graduates who could have attended the flagship University of North Carolina-Chapel Hill or North Carolina State - but instead went to less selective schools - they conclude barely one-third even applied to the state's leading universities. Most of those that applied got in but went elsewhere, or nowhere.

Those students who "undermatched" may have figured they would be in for an easier time; they did in fact get higher grades, but overall paid "a high price," taking longer to move through school and eventually graduating at a rate 15 points lower than comparably prepared students who went to more selective schools.

"We do not mean to suggest that every student should attend the most selective institution for which he or she might qualify," write the authors, who overall looked at 21 flagship public universities and 47 other state institutions. However, students choosing colleges below their qualifications "should not be the norm."

It's well known that more selective colleges generally have higher graduation rates, but the authors say that's not simply because they get better students. They found graduation rates still varied substantially between institutions even when they controlled for academic preparation.

The authors found the explanation wasn't necessarily that selective schools were spending more money per student; rather, it was that they offered a more campus-focused experience. Students who lived in a residence hall their first semester were 7 to 8 percentage points more likely to graduate than those who lived off-campus, even accounting for different academic credentials and background characteristics. That dovetails with other research showing students who make personal connections on campus are more likely to persist.

While the findings might eventually inform students choosing colleges, the more immediate audience is policy-makers and educators. Bowen's previous data-driven work, on affirmative action and college athletics, has been hugely influential. Again with this project, he and his co-authors gained access to information allowing them to track thousands of individual students over time.

The findings paint a grim picture of wasted opportunities, but also suggest even relatively modest efforts to provide students more information and encouragement could substantially "increase social mobility and augment the nation's human capital." UNC is also home to a program of the kind the findings suggest could help boost completion rates - the National College Advising Corps, which places recent college graduates in high schools as college counselors.

Among other conclusions from "Crossing The Finish Line":

- The SAT and ACT standardized college tests are of "exceedingly modest" use in predicting who will graduate from college. The authors don't suggest the tests should necessarily be abandoned, but conclude high school grades, AP exams and subject matter tests are more effective at predicting graduation rates. SAT and ACT scores are most helpful to more selective colleges. Those findings could add momentum to a movement among colleges no longer requiring students to submit SAT or ACT scores.

- Students starting at two-year schools aren't as likely to complete a bachelor's degree as comparably prepared students who start at four-year schools. Those findings cast some doubt on the wisdom of programs in several states, such as New Jersey and Virginia, encouraging students to start at two-year colleges, then transfer. However, the authors urge four-year college to consider accepting more community college transfers; those who make to the transfer stage do surprisingly well, and could help make flagship public universities more economically and racially diverse.