The last several years have seen a steady shift in the rhetoric of college administrators and higher education policy experts, with concerns about the "success" of students once they are in college rivaling, if not quite supplanting, the longstanding emphasis on ensuring "access" to a higher education in the first place. Colleges have paid more and more attention to what they do to retain the students they have, from instituting better educational practices such as learning communities to strengthening advising.

By at least one measure, those efforts may not be working. Data released Thursday by ACT, Inc., show that the proportion of first year students who returned to their colleges as sophomores in 2007-8 -- 65.7 percent -- dropped to the lowest level in the 25 years the organization has been collecting the information. The figure was 68.1 percent in 2006-7 and 68.7 percent -- the highest ever -- just two years ago, in 2005-6.

The 2007-8 numbers represented a low point for five of the eight categories of institutions measured: two-year private colleges (55.5 percent), four-year private baccalaureate colleges (69.6 percent), private master's institutions (72.3 percent) and private and public doctoral granting universities (80.4 and 72.9 percent, respectively).

But the retention rate actually rose for one major set of institutions: two-year public colleges. Community colleges saw their overall retention rate rise to 53.7 percent, up from a low point of 51.3 percent in 2004.

There are many, many caveats to the numbers, as even ACT admits. First, and probably most importantly, the data measure only whether or not students returned to their own institutions for a second year, so students who transferred to another college immediately, or took a year off and then returned, are not differentiated from those who left to drive a cab. That means that the ACT survey does not account for the type of "swirling" educational paths about which Clifford Adelman and other researchers have written, in which students move in and out of individual colleges, and higher education altogether, and yet still move toward their educational goals.

Second, over the 25 years ACT has been cataloging this information, the number of undergraduate students at American colleges has risen sharply, and with that democratization, their academic preparation, at least as measured by traditional standards, has declined. "The students who now have access are very heterogeneous, so therefore, stability [in the general national retention rates over time] is probably success," said Wes Habley, a principal associate at ACT who produces the retention survey.

So with those caveats, are the numbers still troubling? Experts are divided. George D. Kuh, Chancellor's Professor of Higher Education and director of the Center for Postsecondary Research at Indiana University at Bloomington, said he was heartened -- looking at "glass half full" -- by the fact that the retention rates had generally held steady even as so many more students, many of them academically underprepared, flowed into higher education. "In terms of raw numbers, there are many more students" -- millions more, arguably -- "persisting in higher education today than in 1988," he said.

Still, Kuh said it was hard not to be disappointed by the fact that the retention rates appear to be dropping at a time when colleges are focusing more than ever before on student retention and success. "There's obviously a lot of talk about persistence, but the concern is that what's actually being done may not be so effective," Kuh said.

"These numbers ought to sound an alarm bell," he said, "but to understand what's going on, you're going to have to look at the individual campus level."

The bells are not going off for Cliff Adelman, though. The longtime Education Department researcher, now an analyst at the Institute for Higher Education Policy, said the ACT report's inability to say what happened to the 34 percent of freshmen who did not return as sophomores renders the data nearly meaningless. (The three-percentage-point drop, he said, may be statistically significant, but "it's not very meaningful.")

Federal longitudinal data on students in the 1990s finds that "of all those students who started in four-year colleges, only 5 percent did not come back [to college] somewhere and at some time during their scheduled second year," he said in an e-mail message. And of students who started in community colleges, "only 16 percent did not show up somewhere and at some time during their scheduled second year."

Adelman says that rather than focus on the "retention" of students by one college -- which he decries as overly focused on the fates of "schools that want to hang on to students (one suspects, more for the sake of their reputations than what would be best for the student)" -- policy makers would be better off focusing on the rates at which students persist in higher education, which is "much more important to the rest of us who have worked to identify the different productive paths students can take in space and time through our system."

After all, Adelman points out, the new president of the United States would have shown up as an un-retained student in some studies (though not in the ACT data, which examines students after only one year) when he left Occidental College after his sophomore year and landed at Columbia University a few decades ago....