NBER paper by Bound, Lovenheim, and Turner (2009).
Partly as a consequence of the substantial increase in the college wage premium since 1980, a much higher fraction of high school graduates enter college today than they did a quarter century ago. However, the rise in the fraction of high school graduates attending college has not been met by a proportional increase in the fraction who finish. Comparing two cohorts from the high school classes of 1972 and 1992, we show eight-year college completion rates declined nationally, and this decline is most pronounced amongst men beginning college at less-selective public 4-year schools and amongst students starting at community colleges. We decompose the observed changes in completion rates into the component due to changes in the preparedness of entering students and the component due to collegiate characteristics, including type of institution and resources per student. We find that, while both factors play a role, it is the collegiate characteristics that are more important. A central contribution of this analysis is to show the importance of the supply-side of the higher education in explaining changes in college completion.
The emphasis is mine. Not only are men a minority on many college campuses, many aren't finishing when they do start.