Matching and "undermatching" in college admissions

A recent commentary in the Chronicle of Higher Education summarizes a new book on how (in)frequently students finish college in the canonical four years:

Helping Students Finish the 4-Year Run By William G. Bowen, Matthew M. Chingos, and Michael S. McPherson


They make a number of points, two of which focus on matching. They find that many students, particularly from lower socio-economic backgrounds, "undermatch," by going to less selective colleges than they are qualified for. And they think that standardized tests play too large a role in sorting students to colleges.

"5. But money is by no means the entire story, perhaps not even the largest part. Student's choices of where to apply to college are enormously important. A surprisingly large number of students—especially those from poor families and those who are African-American or Hispanic—"undermatch." That is, they go to less demanding four-year institutions than they are qualified to attend, to two-year colleges, or to no college at all. For example, 59 percent of students in the bottom quartile of family income undermatch; 27 percent in the top quartile do so. In addition, 64 percent of students whose parents have no college education undermatch, compared with 41 percent of those whose parents have college degrees and 31 percent whose parents have graduate degrees (see Figure 3). Undermatching has serious consequences because there is a strong association between institutional selectivity and B.A.-completion rates: Students with essentially the same qualifications who attend more-selective universities have a considerably higher probability of graduating than do comparable students who attend less selective universities. Our data also confirm the results of other studies that show that students whose objective is to earn a B.A. are much less likely to do so if they start at a two-year college (again, other things equal).
6. "Sorting" of applicants by universities, especially overreliance on standardized tests, is consequential and problematic. We are not opposed to testing per se. Standardized tests can be helpful when used in the right ways and in the right settings. They are especially helpful when used with high-school grades to predict college grades at the most selective universities. It is clear, however, that high-school grades are far better predictors of graduation rates, especially at less selective universities. This finding holds even when we do not take account of differences in the quality of the high school that a student attended. Results of achievement tests, especially scores on Advanced Placement tests, are also good predictors. Both grades and achievement-test scores measure not only cognitive achievement but also coping and time-management skills—which, we surmise, affect completion rates."