Theda Skocpol and Suzanne Mettler write (free but awkward registration required),

in 1970, 6.2 percent of the U.S. population in the bottom income quartile had completed a baccalaureate degree by age 24-and that percentage actually declined slightly, to 6 percent, by the year 2000. Lower-middle-income young people from the second (to the bottom) income quartile improved their college completion rates only slightly from 1970 to 2000, from 10.9 percent to 12.7 percent. But note the contrasting trajectories for young people in the upper half of the income distribution. For those in the third quartile-solidly middle-class families-completion percentages rose markedly, from 14.9 percent in 1970 to 26.8 percent in 2000. And for the most privileged young people, those from upper-middle-class and upper-class families in the top quarter of the income distribution, college completion rates rose from 40.2 percent in 1970 to 51.3 percent in 2000. Compared to the mid-twentieth century, higher education is now increasingly exacerbating socioeconomic inequality in the United States.

Of course, this all could reflect assortive mating. Today’s upper-income children could be the offspring of two parents who both went to college, while the lower-income children could be the offspring of two parents who did not. Instead of thinking along these lines, the authors favor a big Federal program of college financial aid. The article pretty much assumes that everyone could benefit from college, and the only reason some people don’t go is that they cannot afford it.

Thanks to Timothy Taylor’s “Recommendations for Further Reading” column in the Journal of Economic Perspectives for the pointer.